Open Access
Add to BibSonomy
Abstract
Tissue clearing methods render biological tissues transparent while maintaining tissue structure, enabling visualization of entire tissues. Recent developments in tissue clearing have predominantly emphasized preserving intrinsic fluorescent proteins or aqueous-based tissue clearing and so typically involve complex procedures and long processing times. The utilization of tissue clearing protocols in standard of care histology settings has been less well explored, and protocols for rapid clearing of human tissue specimens are limited. This study presents a novel rapid clearing protocol and demonstrates a low-cost tissue processor for high volume rapid tissue clearing that can be intergraded into standard histology workflow. We demonstrate rapid clearing in dermatological specimens, including both nonmelanoma and melanoma excisions.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Malignant melanoma (MM) or localized malignant melanoma in situ (MMIS) affect close to 200,000 Americans annually [1]. While surgical excision of melanoma prior to metastasis has a high cure rate, the five-year survival rate reduces to only 23% with distant metastasis [1]. As MM and MMIS frequently occur in sun-exposed regions like the face and neck, tissue conservation is critical to avoid complex reconstruction and ensure a good cosmetic outcome [2]. However, MMIS often have irregular borders, and malignant cells can extend beyond the apparent tumor border, which increases the risk of recurrence and the potential for metastasis [3]. Consequently, surgery for early stage melanoma faces a tradeoff between the risks of recurrence and tissue conservation [4], making accurate assessment of tumor extent critical.
Current options for intraoperative histology during MMIS treatment include Mohs micrographic surgery (MMS) with en face frozen sections with rapid (∼1hr) immunohistochemistry (IHC) and staged surgical excision (SSE) with overnight (∼24hr) paraffin sections. In addition to long processing times that extend surgery for hours or days, these techniques rely on sparse en face sections that evaluate only a very limited number of sections within the tumor margin volume. While MMS with IHC and SSE enable more tissue conservation by comprehensive evaluation of the margin surface, both have a limited ability to visualize the true extent of melanoma because the diffuse spread means that isolated planes can easily miss malignant melanocytes. Thus, both methods are not recommended for higher risk diseases for which treatment imposes the highest morbidity. Instead of attempting to detect positive margins from only a limited number of sections as in MMS with IHC or SSE, a volumetric imaging method would potentially allow more sensitive detection of residual disease while reducing the laborious sectioning process through margins.
Recent developments in two photon fluorescent microscopy (TPFM) have demonstrated real-time surface imaging of dermatologic excisions in a point of care setting [5]. TPFM is particularly advantageous for melanocytic tissues because infrared wavelengths are much more weakly absorbed than visible wavelengths. While TPFM provides limited volumetric imaging capabilities into highly scattering tissues, maximum image depths are still limited by scattering and optical aberration induced by nonhomogeneous tissue refractive index [6]. Surface imaging is sufficient for histologic scenarios where only a single section is required but cannot replicate standard step sectioning where serial sections are taken at depths through the volume of a tissue. Tissue clearing is a rapidly developing field that has revolutionized the study of three-dimensional (3D) tissue architecture [7]. Cleared tissue two photon fluorescence microscopy (cTPFM) can potentially enable volumetric histology through entire excisions. Furthermore, new detector technology, such as silicon photomultipliers (SiPM) [8] in conjunction with new scanning method such as scanner synchronized strip scanning [9] enables cTPFM at extremely high imaging rates, making volumetric imaging of large tissue specimens feasible.
Although tissue clearing techniques were first developed more than 100 years ago [10], there has been renewed interest in developing new clearing protocols [11]. Many recent studies have focused on endogenous fluorophore preservation in research settings [12,13] or improved immunofluorescence [14,15]. However, applications in clinical settings for which transgenic proteins are not present have been more limited, especially for highly scattering human skin tissue [16]. Currently, there are three main categories of tissue clearing techniques: solvent based techniques [17,18], hyperhydration techniques such as clear, unobstructed brain imaging cocktails (CUBIC) [14], and hydrogel embedding techniques such as CLARITY [15]. However, both the hyperhydration and hydrogel embedding clearing protocols are extremely time consuming, on the order of days for both CUBIC and CLARITY [14,15]. For example, recent work has explored the CUBIC protocol to clear human melanoma samples but it required several days just for staining and index matching [19].
The fastest methods are generally solvent-based and involve initial dehydration steps that extract water from tissue while labeling nuclei and cytoplasm with DNA labels such as DAPI and counterstains such as eosin and sulforhodamine 101 (SR101) [17,18]. Following dehydration and staining, a high index immersion medium such as 1:2 benzyl alcohol (BA) with benzyl benzoate (BB), dibenzyl ether (DBE), or ethyl cinnamate (ECi) can be diffused into the tissue to match the refractive index of protein and render tissue transparent [18,20]. Unfortunately, while faster than other methods, this process is still time-consuming, with protocols in the literature typically requiring at least 5 hours for needle biopsies to days for larger specimens. Conversely, ultrafast optical clearing methods such as DMSO [21], TDE [22], and nontoxic ultrafast optical clearing method (FOCM) [23] can partially clear tissue in minutes but typically result in only modest reductions in scattering. For example, DMSO and TDE have very limited imaging depth [22], especially in highly scattering tissues such as skin. Similarly, while FOCM can clear in only 2 minutes, the maximum imaging depth in the brain increased only 3-fold. The limited imaging depth using these methods makes them unsatisfactory for imaging thick surgical specimens. Recent work by Olson et al. has attempted to address this limitation by combining faster processing with demonstrated clearing of small biopsy specimens in 6 hours using BABB as a clearing media and DAPI as a nuclear stain [18]. Subsequently, this was improved using elevated temperatures, clearing small biopsies in just 2 hours (0.5hr fixing and 1.5 dehydration and clearing) [20]. While promising, there were some limitations. First, the use of DAPI requires <800 nm ultrafast sources for two photon excitation, which are more costly to integrate into clinical instruments than longer wavelength sources. Second, the protocol was limited to relatively small-sized needle biopsy samples that could be rapidly fixed and dehydrated rather than larger histological specimens which would take longer. In this work, we develop a new rapid tissue clearing protocol using a low-cost tissue processor capable of batch clearing larger, unfixed dermatological specimens within 3.5 hours of excision. This protocol extends the TPFM imaging depth more than 60-fold to over 2.5 mm in human skin. Since cTPFM imaging does not require sectioning, the cleared tissue can be imaged within hours of excision, enabling same-day evaluation and diagnosis. Finally, we show that the protocol is fully compatible with existing histology processes and that all fluorescence labels are washed out during conventional downstream paraffin processing.
2. Material and method
2.1 Nuclei stain evaluation and washout
Stain specificity was tested using a constant 20 µg/ml of each stain on frozen sections made from discarded human skin remaining after surgery. For the SYBR dyes, this equated to 20X concentration (1:500 dilution from stock). Samples were then imaged using TPFM with 1040 nm excitation, and the fluorescence signal was collected from 525-575 nm emission. These were rendered into virtual H&E images using a virtual transillumination algorithm [24].
To evaluate the persistence of each stain following paraffinization, thawed discarded frozen skin tissue samples were stained for 3 minutes with 20µg/ml acridine orange (AO), and thiazole orange (TO); 20X SYBR Gold (SG), SYBR Safe (SS), and SYBR Green (SGr) as a nuclei stain, and 20 µg/ml SR101 in 70% EtOH. These samples were imaged using TPFM with 1040 nm excitation in the whole mount configuration, as described in previous work [5]. Nuclei stain emission was collected between 525-575 nm, SR101 emission was collected between 602-682 nm, and the Second Harmonic Generation (SHG) signal between 501-521 nm as a washout reference. After TPFM imaging, the samples underwent a standard 7.75-hour paraffin tissue processing cycle used by the Department of Pathology at the University of Rochester Medical Center, consisting of various steps involving formalin, ethanol, and xylene, as shown in Fig. 2(F). The resulting paraffin blocks were sectioned into two 5µm sections and mounted onto separate glass microscopy slides. The first section was stained with the traditional H&E stain and then scanned using an Olympus VS120 slide scanner. The second unstained paraffin slide was then deparaffinized and cover-slipped before being imaged using the same excitation and emission collection windows as the whole mount sample. TPFM images before and after the washout were co-registered based on the SHG signal.
2.2 Tissue collection
Human surgical margins remaining after cryosectioning for Mohs surgery or rapid IHC were collected under a protocol approved by the Research Subjects Review Board (RSRB), which serves as the IRB for the University of Rochester. As tissues were collected after the conclusion of diagnostic procedures and without patient contact, the RSRB waived the need for informed consent.
2.3 Clearing protocol evaluation
For testing immersion media, discarded tissues were dehydrated and delipidated using the room temperature protocol in Table 1 within 1 hour of excision. After the dehydration process, the samples were placed in either dibenzyl ether (DBE) or ethyl cinnamate (ECi) for 0.5 hours simultaneously. For testing elevated temperature clearing, thawed discarded frozen skin tissue samples were dehydrated using the elevated temperature protocol in Table 1 and compared to another skin tissue sample dehydrated at room temperature, both for the same duration.
Table 1. Tissue clearing protocol
2.4 White light imaging setup
Images were acquired using a USB webcam (Arducam) placed above a grid target. The NIR images were taken without an IR filter in front of the webcam sensor. Thawed frozen skin and fresh skin tissues were placed on a 1 mm thick glass (3 inch x 4 inch) above a grid target with either visible or near infrared (830 nm) backlighting before clearing. Cleared tissue were suspended inside a clear glass cuvette (LAB4US) filled with corresponding clearing medium. The glass cuvette was placed on the same 1 mm thick glass above a grid target with same white light and NIR backlighting.
2.5 cTPFM imaging
The same system as the nuclear stain evaluation was utilized for cleared tissue imaging. Each tissue sample was mounted inside a modified histology cassette with a clear optical window for imaging and submerged in the clearing medium. All images were acquired using an ASI multi-immersion objective (ASI Valve 54-12-8 0.7 NA) with a custom dipping cap.
2.6 MMIS surgical margin imaging
Frozen tissue was thawed, cleared using the DBE protocol, and then imaged within 2 hours of clearing. After imaging with cTPFM, the tissue was placed into 100% EtOH for 15 mins (to reverse the clearing process) and then placed into formalin for fixation. The fixed MMIS tissues were then submitted to the University of Rochester Medical Center Department of Pathology for MART-1 IHC analysis on the en face margin. The MART-1 slides were then scanned with a Grundium Ocus20 slide scanner.
3. Result
3.1 Nuclei stain evaluation
We selected green fluorescent dyes such as AO [25], TO, and SG [26], which have emission ranges of approximately 530-580 nm and are compatible with compact and inexpensive 1µm excitation sources frequently used for two-photon histological imaging [25]. To expand our evaluation, we also examined other less commonly used SYBR family stains, including SS and SGr. Ensuring uniform staining in large, cleared tissues requires extended staining times because the stain has to diffuse millimeters into the tissue from the surface. However, extended staining time can lead to non-specific staining, necessitating very high specificity to maintain high contrast. Although all of the nuclear stains tested were similarly compatible with 1040 nm two photon excitation and produced comparable brightness, SGr and SG stand out in terms of specificity, as shown in Fig. 1. AO, TO, and SS stained a portion of cell cytoplasm inside of sebaceous glands as well as the target cell nuclei, resulting in reduced contrast. Although widely used for rapid histological imaging [27–29], AO was the least specific among all the stains tested, with substantial staining of collagen and connective tissue. Conversely, SGr and SG had minimal cytoplasm and connective tissue staining, while SS and TO had intermediate non-specific staining.
Fig. 1. Virtual hematoxylin images produced from five green fluorescent DNA labels. A&B: Skin tissue stained with AO. AO stained a wide range of connective tissue around the sebaceous glands and sebocytes’ cytoplasm. C&D: Skin tissue stained with TO. Less connective tissue is stained with TO compared to AO, but the cytoplasm of sebocytes was still stained by TO. E&F: Skin tissue stained with SS. SS has similar specificity as TO, with minor unspecific staining of sebocytes’ cytoplasm. G&H: Skin tissue stained with SGr. SGr showed a very specific stain with no labeling on both connective tissue and cytoplasm. I&J: Skin tissue stained with SG. SG has similar specificity as SGr.
Fig. 2. A: cTPFM image of SGr (red channel) and SR101 (green channel) stained images after 4.5 hours total stain time. The SHG channel (blue) is completely attenuated by the strong SR101 absorption of 520 nm light. B: Corresponding paraffin embedded slide imaged with 1040 nm two photon exaction. Second generation (SHG) is readily visible (blue channel), but there is no residual SGr (red channel) and SR101 (green channel) fluorescence in the red/green channels. C: Same image in B with 10X digital gain. With 10X digital gain, the week SHG signal has saturated, but no SGr or SR101 is detectable. D: TPFM image of hawed human skin tissue labeled with SG showing strong SHG and nuclei signal. E: TPFM image of corresponding paraffin embedded slide at around the same region as D with same gain value. SG showed extremely strong staining even after paraffinization. Due to the tilted sectioning plane, the same nuclei cannot be precisely co-registered between D and E. F: Paraffin tissue processing cycle used by the Department of Pathology at the University of Rochester Medical Center.
We next assessed the compatibility of each stain with conventional paraffin-based tissue processing. While interference with H&E is extremely unlikely given the high concentration of hematoxylin and eosin used in commercial tissue processors, other widely used diagnostics such as IHC [30] and fluorescence in situ hybridization (FISH) may be more susceptible to interference. Thus, it is preferable if fluorescent stains do not persist through paraffinization. We submitted fluorescence-stained tissues for standard paraffin processing and embedding at the University of Rochester Medical Center. Following processing, TPFM images of slides showed only SHG with the complete elimination of fluorescence from AO, TO, SS, and SGr. However, SG showed extremely strong fluorescence even after paraffinization, as shown in Fig. 2 D vs E, indicating that it could not be removed by conventional tissue processing using the protocol shown in Fig. 2(F). Finally, we replicated these measurements after dehydration and 4.5 hours of staining time and index matching to clear the tissue using the room temperature protocol in Table 1. We confirmed that the longer staining time and clearing process still enabled the complete extraction of residual dye by routine paraffinization (representative images in Fig. 2).
3.2 Low-cost automated tissue processor
To enable high-throughput and reproducible clearing of tissues, we developed an automated tissue processor that can perform the clearing protocol autonomously, ensuring consistency and enabling elevated temperature during the dehydration and delipidation processes as shown in Fig. 3. The processor features a dedicated peristaltic pump for up to 6 working solutions (typically 70%, 95%, 100% EtOH, xylene, and a clearing media such as BABB or DBE), controlled by an Arduino Uno (Arduino, LLC) to transfer the chemical into the tissue processing chamber. After each step, an additional pump extracts the chemical and sends it to a waste container. A heating element (50W 6ohm load resistor), controlled with a PID controller (W1209), maintains a set temperature (typically 40C). The operation of the automated tissue clearing system was validated by successfully clearing tissue using an established BABB protocol.
Fig. 3. Low-cost tissue automated tissue processor for room temperature and accelerated elevated temperature clearing. A: Overall design scheme of the automated tissue processor. B: Assembled low-cost automated tissue processor.
3.3 Selection of index matching media
Common index matching media used in tissue clearing include a 1:2 benzyl alcohol (BA) and benzyl benzoate (BB) mixture, DBE, and ECi. Although the BABB mixture provides good index matching, it has poor compatibility with nuclei stains, resulting in gradual washout of some stains due to solubility in BA. Furthermore, we observed focus drift during imaging, suggesting that the clearing media slightly absorbed 1040 nm illumination and was thermally expanding. Conversely, both DBE and ECi preserved nuclei fluorescence, were thermally stable, and exhibited good index matching in the visible spectrum, as shown in Fig. 4. However, ECi-cleared tissues have slightly more scattering than DBE-cleared tissues in the NIR spectrum, as shown in the NIR transmission image in Fig. 4(E)&(F).
Fig. 4. ECi and DBE immersion medium index matching comparison using room temperature dehydration protocol with 4.5 hours staining and dehydration in gradient EtOH. A&B: Freshly excised human skin tissue in PBS. C: Cleared human skin tissue in DBE under visible backlight. D: Cleared human skin tissue in ECi under visible backlight. E: Cleared human skin tissue in DBE under NIR LED backlight. F: Cleared human skin tissue in ECi under NIR backlight.
3.4 Clearing protocol optimization
Using the tissue processor to accelerate testing larger numbers of specimens, frozen skin tissues, and fresh nonmelanoma skin cancer tumor cores were collected. Frozen skin tissues were thawed before dehydration, while fresh NMSC tumor cores were dehydrated within one hour of excision. Skin specimens were dehydrated via a graduated EtOH stain assay, consisting of 70% EtOH (SGr 20X; SR 10 µg/ml) for 2 hours, 95% EtOH (SGr 20X; SR 10 µg/ml) for 1.5 hours, 100% EtOH (SGr 20X; SR 5µg/ml) for 0.5 hours, and 100% EtOH (SGr 20X; SR 5 µg/ml) for another 0.5 hours. The specimens were then immersed in xylene (0.5 hours) for delipidation, followed by index matching. We did not further optimize BABB due to the turbulence produced by 1040 nm absorption.
To determine if temperature affects the clearing process, we tested dehydration at 20C and 40C and found that both produced comparable results. Next, we iterated on the dehydration time, iteratively lowering it until the tissue was no longer cleared. As a result, the total dehydration time was reduced from 4.5 hours at room temperature to 2.75 hours with elevated temperature EtOH dehydration, with each stage of dehydration taking approximately half as long at 40C as at 20C (Table 1 and Fig. 5). Due to the relatively short time and safety concern for both xylene and index matching media, we did not test elevated temperature for these steps. We conclude that this accelerated protocol obtained similar results but required only 60% of the clearing time. The final clearing protocols in Table 1 are very similar to standard protocols for paraffin embedding tissue, suggesting ease of integration into current pathology labs by simply adding SGr and SR101 into dehydration steps, removing the tissue before paraffin infusion, and placing it into DBE.
Fig. 5. Accelerated tissue clearing protocol with elevated temperature dehydration process. A&B: Thawed human skin tissue in PBS. C: Partially cleared human skin tissue with a 2.75 dehydration cycle at room temperature suspended in DBE. D: Cleared human skin with 2.75 hours of elevated temperature dehydration. E: Partially cleared human skin tissue with 2.75 room temperature dehydration cycle under NIR backlight. F: Cleared human skin with 2.75 hours elevated temperature dehydration cycle under NIR backlight.
3.5 Rapid clearing of unfixed human tumors
Next, we used the optimized protocol to clear fresh, unfixed, and highly scattering human basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) tumor cores. Following clearing, the specimens were imaged using TPFM. We found that specimens could be easily imaged to a depth of over 2 mm (Fig. 6), typically through the entire excision. Figure 7 shows the SCC tumor core with noncancerous cells on the surface, as shown in Fig. 7(A), B. The tumor tip is not visible on the excision surface layer until 300 µm from the surface, as shown in Fig. 7(C). As the image depth is increased, the tumor tip becomes visible, and the image stack shows that the tumor is connected as indicated by the arrows in Fig. 7(C)-(E). The image stack fully images through the tumor volume, including a major hair follicle at the center.
Fig. 6. cTPFM image of a cleared BCC tumor core with virtual H&E rendering over 0-2.5 mm depth. A: BCC tumor core image at the surface of the tumor excision with clusters of malignant basal cells. The image stack starts from the deepest skin layer (dermis) to the top of the skin layer (epidermis). This tumor excision s B: BCC tumor core image at 0.625 mm depth. The malignant basal cell clusters become larger as they approach the epidermis. C: BCC tumor core image at 1.25 mm depth with progressively larger malignant basal cell cluster. D: BCC tumor core image at 1.875 mm depth. The smaller clusters in A-C are merged into one big cluster as the imaging plane reaches close to the basal layer of the epidermis. E: BCC tumor core imaged at 2.5 mm depth. The image plane has passed through the whole excision and reached the stratum corneum layer. F: 3D volumetric rendering of the BCC tumor core in A-E (0-2.5 mm). The dataset contains 500 layers that are 5µm apart.
Fig. 7. cTPFM image of a cleared SCC tumor core with virtual H&E rendered at 0-1300 µm depth. A: Tumor core surface with no visible SCC tumor cells. B: 150µm from tumor core surface with no visible SCC tumor cells. C: 300 µm from tumor core surface with small tumor tip of SCC tumor visible as indicated with green arrow. D: 450 µm from tumor core surface with different tips of solid tumor core indicated with arrows. E: 600 µm from tumor core surface, isolated tumor tips join together to form a single SCC solid tumor core. F: 750 µm from tumor core surface with single hair follicle showing in the middle. G&H: 900 µm – 1050 µm from tumor core surface with imaging plane progressively close to tumor core outer surface. I: 3D volumetric rendering of the SCC tumor core in A-H (0-1.42 mm). The dataset contains 141 layers that are 10µm apart. The green and red arrow indicates the region in D&E indicated by the same color arrow.
3.6 Volumetric imaging of MMIS surgical margin
As the detection of residual melanoma poses a significant clinical challenge, we next explored clearing of melanoma in situ surgical margins. Residual margin tissues discarded after surgery were collected (n = 17). The frozen tissue was thawed, cleared using the accelerated DBE protocol, and then imaged within 2 hours of clearing (Fig. 8).
Fig. 8. Cleared MMIS surgical margin at 350 µm depth from the tumor margin. A: Full layer mosaic of MMIS surgical margin. B: Magnified epidermis region boxed in A, with potential melanocytes indicated by arrows. C: The boxed region in B, with arrows indicating the same potential melanocytes in B.
After imaging with cTPFM, the 3 MMIS margins were placed into 100% EtOH for 15 mins to reverse clearing and then placed into formalin for fixation. The fixed MMIS tissue was then submitted to the University of Rochester Medical Center Department of Histology for standard MART-1 IHC analysis. The MART-1 slides were scanned with Grundium OCUS 20X (Fig. 9). Cleared samples showed normal MART-1 staining of melanocytes, similar to never-cleared samples.
Fig. 9. Downstream IHC staining (Mart-1) of cleared surgical margins. The DBE tissue clearing protocol is compatible with downstream IHC analysis. A: Mart-1 IHC staining of a DBE cleared tissue. The cleared tissue is placed in 100% EtOH for 15 minutes before formalin fixation. While sparse individual melanocytes are labeled, the Mart-1 staining suggests this sample is clear of MMIS. B: Full layer mosaic of MMIS surgical margin. C: Boxed region in A shows several mart-1 labeled melanocytes in the basal layer of the epidermis. D: The boxed region in B with a circle indicates potential melanocytes in the circled region in C.
4. Discussion
Histological evaluation of many cancers, including melanoma, depends on step sectioning at intervals through the tissue of interest to evaluate the internal volume for the presence or depth of malignancy. This process is laborious, costly, time-consuming, and subject to sampling error if pathologic regions fall between sections. We report a tissue clearing, staining, and imaging protocol that is much faster than conventional paraffinization and step sectioning while using only inexpensive reagents. We combine this with hardware to completely automate processing, eliminating the extensive labor required to step section through tissue. Following clearing, TPFM can image an arbitrary number of sections within the tissue volume, while virtual H&E rendering can be used to generate images that closely resemble the appearance of paraffin sections. In previous work, we demonstrated TPFM mosaic imaging at a rate of 47 seconds per square cm at a resolution equivalent to a 40x slide scanner [9]. At this imaging rate, typical step sectioning protocols could be reproduced using TPFM in just a few minutes per sample without requiring extensive manipulation by a skilled histology technician.
Acceptance of tissue clearing into histological workflows requires a demonstration of compatibility with existing assays. Toward this, we screened a series of DNA labels for compatibility with solvent-based clearing and extraction by ethanol/xylene as used during paraffinization in clinical histopathology. While our screen reproduced existing results, such as acridine orange being incompatible with solvent-based clearing but extracted by ethanol, it also turned up the surprising fact that SYBR Gold, which has been widely used in previous fluorescence histology imaging studies [26], was extremely resistant to extraction by typical histology solvents. Further, SYBR Gold showed almost no reduction in brightness after paraffinization. Conversely, we identified the closely related SYBR Green as a promising alternative due to its similar appearance and high specificity but complete extraction from tissue. Thus, SYBR Green may be a safer alternative to SYBR Gold on clinical specimens that require subsequent assays. Also of note, the classic stain thiazole orange was less specific than the SYBR dyes, but much less expensive and was also extracted during paraffinization. All stains were compatible with relatively inexpensive Ytterbium laser excitation.
In comparison to other tissues such as the brain, skin is highly scattering, requiring high power and long integration time to image even 100 microns depth using 1040 nm excitation. Similarly, clearing skin is difficult, and many protocols developed for brain and other organs are less effective or even totally ineffective at clearing skin. In this work, we demonstrate a fast (3.25 hours from excision) protocol to increase the maximum imaging depth by ∼2 orders of magnitude by rendering large specimens completely transparent to NIR light. Our protocol was motivated by previous work [20] rapidly clearing small biopsy specimens using elevated temperature. However, to enable deep imaging in highly scattering skin, we add an additional xylene step which very rapidly extracts most lipids from tissue, resulting in much lower scattering. We also evaluated additional index matching media and found that DBE enabled much deeper imaging and was more thermally stable under 1040 nm excitation.
We opted to limit the maximum temperature to 40C and kept xylene at ambient temperature for safety reasons when working with flammable solvents. In testing, we found that a 20C increase in temperature approximately halved the time required at each stage of clearing. However, during histological processing, paraffinization is performed at up to 60C. Thus, our protocol is conservative relative to typical tissue processing and could probably be further accelerated either by further increasing temperature or using elevated temperature for more of the steps. Nonetheless, the few hours of clearing time is dramatically faster than conventional paraffinization and sectioning.
Step sectioning is well established in many areas of dermatology, including melanoma primary excisions [31] and sentinel lymph nodes, where in the latter case, step sectioning through the specimen finds significantly more metastasis [32]. Even in basal or squamous cell carcinomas, where step sectioning is not routinely used, one study found that 20% of biopsies judged benign on initial sections were converted to basal or squamous cell carcinoma on subsequent step sectioning [33]. Thus, the ability shown here to rapidly image and then visualize large numbers of sections without the extreme labor required for step sectioning and the lengthy process of examining large numbers of slides may enable improvements in sensitivity.
5. Conclusion
We have demonstrated an optimized tissue clearing protocol that can clear all common types of dermatologic excisions in a fraction of the time required for paraffinization and with dramatically less manual labor than required for microtome sectioning. Our protocol enables comprehensive evaluation of an arbitrary number of levels into specimens and is compatible with downstream paraffin processing and IHC. These results suggest a new approach to dermatopathology that would be less labor-intensive while potentially improving sensitivity.
Funding
National Institutes of Health (R21-EB032839, R37-CA258376).
Acknowledgments
We thank Beth Geer for helping to collect discarded human skin tissue.
Disclosures
The authors declare no conflicts of interest.
Data availability
The dataset used and/or analyzed during the study are available from the corresponding author on reasonable request.
References
1. R. L. Siegel, K. D. Miller, H. E. Fuchs, et al., “Cancer statistics, 2022,” Ca-Cancer J. Clin. 72(1), 7–33 (2022). [CrossRef]
2. S. M. Stricklin, W. V. Stoecker, J. M. Malters, et al., “Melanoma in situ in a private practice setting 2005 through 2009: Location, lesion size, lack of concern,” J. Am. Acad. Dermatol. 67(3), e105–e109 (2012). [CrossRef]
3. G. M. Bricca, D. G. Brodland, D. Ren, et al., “Cutaneous head and neck melanoma treated with Mohs micrographic surgery,” J. Am. Acad. Dermatol. 52(1), 92–100 (2005). [CrossRef]
4. C. J. Miller, T. M. Shin, J. F. Sobanko, et al., “Risk factors for positive or equivocal margins after wide local excision of 1345 cutaneous melanomas,” J. Am. Acad. Dermatol. 77(2), 333–340.e1 (2017). [CrossRef]
5. V. D. Ching-Roa, C. Z. Huang, S. F. Ibrahim, et al., “Real-time analysis of skin biopsy specimens with 2-photon fluorescence microscopy,” JAMA Dermatol. 158(10), 1175 (2022). [CrossRef]
6. M. Balu, C. B. Zachary, R. M. Harris, et al., “In Vivo Multiphoton Microscopy of Basal Cell Carcinoma,” JAMA Dermatol. 151(10), 1068 (2015). [CrossRef]
7. T. Tian, Z. Yang, and X. Li, “Tissue clearing technique: Recent progress and biomedical applications,” J. Anat. 238(2), 489–507 (2021). [CrossRef]
8. V. D. Ching-Roa, E. M. Olson, S. F. Ibrahim, et al., “Ultrahigh-speed point scanning two-photon microscopy using high dynamic range silicon photomultipliers,” Sci. Rep. 11(1), 5248 (2021). [CrossRef]
9. C. Huang, V. Ching-Roa, Y. Liu, et al., “High-speed mosaic imaging using scanner-synchronized stage position sampling,” J. Biomed. Opt. 27(01), 016502 (2022). [CrossRef]
10. Y. Zhan, H. Wu, L. Liu, et al., “Organic solvent-based tissue clearing techniques and their applications,” J. Biophotonics 14(6), e202000413 (2021). [CrossRef]
11. V. V. Tuchin, D. Zhu, and G. A. Elian, Handbook of Tissue Optical Clearing: New Prospects in Optical Imaging (CRC Press, Boca Raton, 2022).
12. Q.-H. Shan, X.-Y. Qin, N. Zhou, et al., “A method for ultrafast tissue clearing that preserves fluorescence for multimodal and longitudinal brain imaging,” BMC Biol. 20(1), 77 (2022). [CrossRef]
13. D. S. Richardson and J. W. Lichtman, “Clarifying Tissue Clearing,” Cell 162(2), 246–257 (2015). [CrossRef]
14. K. Matsumoto, T. T. Mitani, S. A. Horiguchi, et al., “Advanced CUBIC tissue clearing for whole-organ cell profiling,” Nat. Protoc. 14(12), 3506–3537 (2019). [CrossRef]
15. K. Chung, J. Wallace, S.-Y. Kim, et al., “Structural and molecular interrogation of intact biological systems,” Nature 497(7449), 332–337 (2013). [CrossRef]
16. A. Ertürk, K. Becker, N. Jährling, et al., “Three-dimensional imaging of solvent-cleared organs using 3DISCO,” Nat. Protoc. 7(11), 1983–1995 (2012). [CrossRef]
17. N. P. Reder, A. K. Glaser, E. F. McCarty, et al., “Open-Top Light-Sheet Microscopy Image Atlas of Prostate Core Needle Biopsies,” Arch. Pathol. Lab. Med. 143(9), 1069–1075 (2019). [CrossRef]
18. E. Olson, M. J. Levene, and R. Torres, “Multiphoton microscopy with clearing for three dimensional histology of kidney biopsies,” Biomed. Opt. Express 7(8), 3089–3096 (2016). [CrossRef]
19. V. Llorente, D. Sanderson, A. Martín-Gorgojo, et al., “A 3D Analysis of Cleared Human Melanoma,” Biomedicines 10(7), 1580 (2022). [CrossRef]
20. R. Torres, E. Olson, R. Homer, et al., “Initial Evaluation of Rapid, Direct-to-Digital Prostate Biopsy Pathology,” Archives of Pathol. & Laboratory Med. 145(5), 583–591 (2021). [CrossRef]
21. A. K. Bui, R. A. McClure, J. Chang, et al., “Revisiting optical clearing with dimethyl sulfoxide (DMSO),” Lasers Surg. Med. 41(2), 142–148 (2009). [CrossRef]
22. Y. Aoyagi, R. Kawakami, H. Osanai, et al., “A Rapid Optical Clearing Protocol Using 2,2′-Thiodiethanol for Microscopic Observation of Fixed Mouse Brain,” PLoS One 10(1), e0116280 (2015). [CrossRef]
23. X. Zhu, L. Huang, Y. Zheng, et al., “Ultrafast optical clearing method for three-dimensional imaging with cellular resolution,” Proc. Natl. Acad. Sci. 116(23), 11480–11489 (2019). [CrossRef]
24. M. G. Giacomelli, L. Husvogt, H. Vardeh, et al., “Virtual Hematoxylin and Eosin Transillumination Microscopy Using Epi-Fluorescence Imaging,” PLoS One 11(8), e0159337 (2016). [CrossRef]
25. L. C. Cahill, M. G. Giacomelli, T. Yoshitake, et al., “Rapid virtual hematoxylin and eosin histology of breast tissue specimens using a compact fluorescence nonlinear microscope,” Lab. Invest. 98(1), 150–160 (2018). [CrossRef]
26. Y. Chen, W. Xie, A. K. Glaser, et al., “Rapid pathology of lumpectomy margins with open-top light-sheet (OTLS) microscopy,” Biomed. Opt. Express 10(3), 1257–1272 (2019). [CrossRef]
27. Y. K. Tao, D. Shen, Y. Sheikine, et al., “Assessment of breast pathologies using nonlinear microscopy,” Proc. Natl. Acad. Sci. 111(43), 15304–15309 (2014). [CrossRef]
28. M. G. Giacomelli, T. Yoshitake, L. C. Cahill, et al., “Multiscale nonlinear microscopy and widefield white light imaging enables rapid histological imaging of surgical specimen margins,” Biomed. Opt. Express 9(5), 2457–2475 (2018). [CrossRef]
29. D. Gareau, A. Bar, N. Snaveley, et al., “Tri-modal confocal mosaics detect residual invasive squamous cell carcinoma in Mohs surgical excisions,” J. Biomed. Opt. 17(6), 066018 (2012). [CrossRef]
30. N. Bouzari and S. Olbricht, “Histologic pitfalls in the Mohs technique,” Dermatol Clin 29(2), 261–272 (2011). [CrossRef]
31. G. H. Pitman, A. W. Kopf, R. S. Bart, et al., “Treatment of lentigo maligna and lentigo maligna melanoma,” J Dermatol Surg Oncol 5(9), 727–737 (1979). [CrossRef]
32. R. Riber-Hansen, N. Hastrup, O. Clemmensen, et al., “Treatment influencing down-staging in EORTC Melanoma Group sentinel node histological protocol compared with complete step-sectioning: a national multicentre study,” Eur J Cancer 48(3), 347–352 (2012). [CrossRef]
33. H. R. Carag, V. G. Prieto, L. S. Yballe, et al., “Utility of Step Sections,” Arch. Dermatol. 136(4), 471–475 (2000). [CrossRef]
