|Home | About | Journals | Submit | Contact Us | Français|
Hypertrophic scars (HTS), frequently seen after traumatic injuries and surgery, remain a major clinical challenge due to the limited success of existing therapies. A significant obstacle to understanding HTS etiology is the lack of tools to monitor scar remodeling longitudinally and non-invasively. We present an in vivo, label-free technique using polarization-sensitive optical frequency domain imaging (PS-OFDI) for the 3D, longitudinal assessment of collagen remodeling in murine HTS. In this study, HTS was induced with a mechanical tension device for 4 to 10 days on incisional wounds and imaged up to one month after device removal; an excisional HTS model was also imaged at 6 months after injury to investigate deeper and more mature scars. We showed that local retardation (LR) and degree of polarization (DOP) provide a robust signature for HTS. Compared to normal skin with heterogeneous LR and low DOP, HTS was characterized by an initially low LR, which increased as collagen fibers remodeled, and a persistently high DOP. This study demonstrates that PS-OFDI offers a powerful tool to gain significant biological insights into HTS remodeling by enabling longitudinal assessment of collagen in vivo, which is critical to elucidating HTS etiology and developing more effective HTS therapies.
Hypertrophic scars (HTS) are conspicuous, potentially disfiguring lesions that remain a major therapeutic challenge. In fact, its incidence rate following surgery in burn injury is particularly high (>90%) (Gauglitz et al., 2011). Apart from the cosmetic disfigurement, HTS frequently recur and can cause pruritus, pain, functional impairment, and profound psychological effects.
The pathogenesis of HTS is not fully understood. Its formation typically involves an abnormal wound healing response to trauma, such as burns, inflammation, or surgery, especially when the wound crosses joints or skin creases at right angles (Wolfram et al., 2009). Among other factors, tension is known to be critical to the formation of HTS and indeed, the well-established surgical treatment – Z-plasty or W-plasty (English and Shenefelt, 1999) – works by relieving tension along the scar. Decreased cellular apoptosis was observed in murine models in which mechanical stress was applied to healing wounds (Aarabi et al., 2007). In addition, overproduction of TGF-β and PDGF suggest the pathologic persistence of wound healing signals (TGF-β1 and β2 stimulate collagen synthesis) in HTS (Gauglitz et al., 2011). Histologically, hypertrophic scars are characterized by the presence of nodular structures with fine, randomly organized collagen bundles (Ehrlich et al., 1994).
To understand scar etiology and assess treatment outcome, imaging is needed. Several imaging techniques have been investigated to assess collagen ex vivo and in vivo. Historically, picrosirius red staining with polarization microscopy, which enhances tissue birefringence (Junqueira et al., 1979), was used to visualize collagen ex vivo. More recently, second-harmonic generation and two-photon excited fluorescence techniques have been used to analyze collagen and elastin, respectively, in ex vivo HTS and keloid specimens (Chen et al., 2009; Vogler et al., 2011). Unfortunately, these techniques only offer a limited field of view and are challenging to use in vivo. Optical coherence tomography (OCT) offers fast imaging speed, increased field of view and imaging depth, at a spatial resolution suitable for biological applications, which makes it particularly amenable to in vivo imaging. OCT has been investigated for the imaging of skin (Alex et al., 2010; Gambichler et al., 2011), and in particular for the imaging of burn scars (Gong et al., 2013) and skin cancer (Mogensen et al., 2009).
Polarization-sensitive OCT (PS-OCT) is an extension of OCT that also measures the polarization state of the light backscattered by the sample. It enables the measurement of tissue birefringence, primarily caused by fibrillar collagen, in the skin. PS-OCT has been shown to provide intrinsic contrast in thermally damaged tissue, thereby providing a tool for burn depth assessment (De Boer et al., 1998; Park et al., 2001; Pierce et al., 2004a, 2004b) and mapping of dermal birefringence in photoaged skin (Sakai et al., 2008) in vivo.
In this study, we present the use of polarization-sensitive optical frequency domain imaging (PS-OFDI), a variant of PS-OCT which offers improved imaging speed and sensitivity (Villiger et al., 2013), for the 3D imaging of a rat model of surgical HTS in vivo. This animal model employs a mechanical tension device placed on healing incisional wounds, which enables the systematic study of HTS formation (Aarabi et al., 2007). To validate PS-OFDI in imaging deeper scars, we also investigated an excisional wound scar placed under autologous tension, imaged at 6 months after injury. We evaluate both local retardation (LR) and degree of polarization (DOP), computed using our spectral binning algorithm (Villiger et al., 2013), for enhancing the contrast of HTS, which, to the best of our knowledge, has not been reported previously. Compared to cumulative retardation, LR expresses the rate of change of the measured polarization states with depth, and is a direct, more intuitive measure of tissue birefringence. DOP is a quantity related to the randomness of the detected polarization states of light, which scales from zero for completely random to unity for perfectly uniform polarization states. Unlike local retardation, DOP captures the cumulative effect of the tissue from the surface to a given depth. We observed that the combination of LR and DOP provides a robust optical signature to differentiate HTS given the presence of fine, densely packed collagen bundles in HTS compared to thicker collagen bundles in normal skin.
In addition, we demonstrate the use of PS-OFDI for studying HTS remodeling by imaging the incisional wound model longitudinally for 1 month and the excisional model at 6 months. Interestingly, we observed a progressive increase in LR and a persistently high DOP within the scar region over time, which corresponded well with the remodeling and thickening of collagen fibers histologically. Normalization of the scar was associated with an increased LR and decreased DOP back to baseline levels in normal skin. Our findings suggest that PS-OFDI can serve as a valuable 3D imaging tool for the non-invasive, longitudinal assessment of HTS in vivo, by providing significant biological insights into collagen remodeling central to understanding HTS etiology and monitoring therapeutic response to improve current therapies and investigate novel approaches.
Figure 1 shows the cross-sectional PS-OFDI images of a mechanical tension-induced hypertrophic scar (HTS) after 10 days of loading (imaged 1 month later) compared to normal rat skin. For intuitive visualization, the LR and DOP signals were merged with the intensity image (as a brightness channel) and displayed using an isoluminant colormap (Geissbuehler and Lasser, 2013). While conventional intensity images show minimal contrast (only slightly increased back-scattering), the local retardation (PS-LR) and degree of polarization (PS-DOP) images demonstrate significant differences in HTS compared to normal skin. HTS shows significantly reduced LR and increased DOP deep inside the dermis, compared to heterogeneous LR and low DOP beyond the epidermis in normal skin. This optical signature (reduced LR and high DOP) corresponds histologically to thin, less organized collagen bundles in HTS (Fig. 1d), compared to thicker collagen bundles in normal skin (Fig. 1h). These abnormal collagen bundles are cigar-shaped and orientated parallel to the surface of the skin along the tension lines of the scar tissue (Fig. 1d). In addition, there is an increased cellularity of fibroblasts in HTS compared to normal skin (Fig.1d and 1h).
We further analyzed the histological correlation of the PS-OFDI images in each animal group with varying duration of tension (Fig. 2). In all cases, the scar region shows reduced LR and increased DOP. Overall, the PS-DOP images correlate well with the extent and shape of the scar as confirmed by H&E histology (Fig. 2c, f, i, l), while the PS-LR images show more variability. The size of HTS also increased with the duration of tension, as expected, from a barely noticeable scar with minimal deposition of collagen in the 4-day group (Fig. 2c) to a significantly larger scar extending all the way through the dermis that is characterized by aberrant collagen bundles and increased cellularity of dermal fibroblasts in the 10-day group (Fig. 2l).
A major advantage of using PS-OFDI, compared to conventional histology processing, is the ability to assess HTS longitudinally and comprehensively in vivo (Fig. 3), which is particularly important for studying HTS etiology and assessing response. By imaging the incisional HTS model (6-day group) at 1-week intervals post device removal, we observed rapid contraction of the scar in the first week, as indicated by the normalization of DOP and LR around the boundary of the scar to baseline levels in normal skin (increased LR and decreased DOP). From weeks 1 to 4, the scar continued to remodel progressively, leading to further reduction in scar size and an interesting increase in LR, particularly in deeper regions. The DOP remained persistently high within the scar region. To investigate the evolution of the LR and DOP signals further, we analyzed the PS-LR and PS-DOP images at three major time points (Fig. 4). After the initial incision (and prior to the application of tension), the fresh incisional wound (at day 2) was marked by a small region with very low LR and high DOP (Fig. 4 a, b), which expanded significantly after loading the healing incision for 8 days (Fig. 4c, d). As the tension-induced wound continued to remodel over the 1-month period, LR increased significantly while DOP remained high (Fig. 4e, f). Finally, we analyzed the relative maturity of the collagen using Herovici’s method (Herovici, 1963), which has been shown and used to distinguish young, newly formed collagen (blue) from more mature, highly cross-linked collagen (purple/red) in previous studies (Krötzsch-Gómez et al., 1998; Lillie et al., 1980; Ozog et al., 2013; Turner et al., 2013). As shown by Herovici’s staining and Ki67/SMA staining, the change in LR over the 1-month period corresponded well with the transition from a scar with thin, newly formed (blue) collagen and myofibroblasts at week 0 to thicker, more mature (purple) collagen bundles with decreased cellularity at week 4 (Fig. 5).
To demonstrate the clinical utility of our approach in imaging deeper full-thickness scars, we performed our imaging experiments in an excisional wound 6 months after the injury. Interestingly, we observed a much more prominent increase in LR deep in the dermis at 6 months (Fig. 6a), which corresponds well with more organized, thicker collagen fibers deep in the dermis with decreased cellularity (Fig. 6e), compared to thinner and less organized collagen fibers surrounded by a high density of fibroblasts in the HTS region closer to the epidermis with a lower LR (Fig. 6d). In both cases, the DOP within the scar remained significantly higher than in normal skin (Fig. 6b).
This study demonstrated the use of PS-OFDI for 3D, label-free imaging of murine models of HTS, thereby enabling non-invasive, quantitative evaluation of scar remodeling longitudinally. Interestingly, we observed that the remodeling of collagen is correlated with a significant increase in LR in both the incisional and excisional HTS models. This ability to probe HTS in vivo allows us to gain significant biological insights into collagen remodeling, which plays a central role in wound healing. While the excisional wound model provides a convenient approach to study deeper scars which are more difficult to treat clinically, the incisional wound model with a tension device provides an elegant way to control the size of the scar systematically (e.g., by varying duration of tension placement) as shown here.
Unlike previous studies employing PS-OCT for the characterization of skin and scar tissue (De Boer et al., 1998; Park et al., 2001; Pierce et al., 2004a, 2004b; Sakai et al., 2008), we reconstructed the local retardation (instead of cumulative retardation), which reflects tissue birefringence more closely and is much more intuitive to interpret. Local retardation expresses how much the birefringence of the tissue rotates the polarization states at each depth, while cumulative retardation compares the polarization states at a given depth with respect to the surface (resulting in increasing values even in uniform tissue with constant birefringence). In addition, we computed DOP, which provides a useful metric to identify and measure HTS. Our expression of DOP is similar to the degree of polarization uniformity established by Gotzinger et al (Götzinger et al., 2008); however, here we employ a spectral binning algorithm and spatial averaging to reduce noise (Villiger 2013). DOP captures the uniformity of the measured polarization states in a small region of interest. In a homogeneous tissue, the polarization states of neighboring pixels are expected to be very similar. In heterogeneous tissues, where two adjacent image pixels potentially interrogate two differently oriented collagen bundles, they are more likely to differ and reduce the measured DOP.
Using LR and DOP, we were able to clearly differentiate normal skin from HTS due to the very different collagen architecture in each case. In normal skin, the heterogeneous, mesh-like organization of highly organized collagen bundles give rise to islands of high dermal birefringence in PS-LR images, alternating with areas of low birefringence due to the varying alignment of the bundles with respect to the optical imaging axis. Furthermore, the spatial dimensions of the region of interest required for the generation of the PS-LR and PS-DOP images frequently include several bundles, accentuating the heterogeneous appearance of normal skin. This heterogeneity also explains the rapid decrease of DOP beyond the epidermis. In contrast, HTS are characterized by finer, and microscopically less organized collagen that exhibits lower dermal birefringence, increasing over time as the fibers remodel and thicken. HTS are optically more homogenous and preserve the polarization state of the incident light, resulting in high DOP values deep within the dermis. In addition, the epidermis and papillary dermis typically showed reduced local retardation values and also preserved the polarization state of the incident light, leading to a high DOP.
In our validation study, the HTS area measured with PS-OFDI showed the expected dependence with the duration of tension and matched well with histological assessment of the same region. DOP provides a measure of the HTS area in the dermis and LR tends to be more variable particularly as the collagen fibers mature. In contrast, DOP typically remains high throughout scar remodeling. One limitation of DOP is that superficial layers normally exhibit high DOP values, which can make it difficult to assess the lateral extent of the scar on the surface.
Our histological analysis confirmed that there is significant hypertrophic scar formation after placing the device for 8 to 10 days. Although the area of scar formation was increased in the 4-day and 6-day groups compared to the control group, histological features of hypertrophic scar were not as evident. This is consistent with an earlier report suggesting that the duration of tension was critical to HTS formation and that at least 7 days of continuous loading was required in a mouse model (Aarabi et al., 2007). It is fascinating that our rat model showed not only significantly increased area of scar in the 10-day group, but also similar histopathological features of HTS in human tissue including increased cellularity of fibroblasts and a dermal nodule with haphazard orientation of collagen bundles.
Finally, this study demonstrated 3D, longitudinal assessment of an HTS model in vivo at high-resolution. We observed rapid remodeling of the scar in the first week after tension device removal, followed by a more progressive remodeling phase, including the emergence of a region with increased birefringence but high DOP deep within the scar. Interestingly, the increase in local retardation corresponded well with the transition from a highly cellular, immature scar with thin collagen fibers at week 0 to a less cellular, more mature scar with thicker collagen fibers at week 4. Our excisional HTS study at 6 months further suggests that LR can be used as a surrogate for collagen remodeling, while DOP can be used to monitor the size of HTS during the remodeling process.
Currently, there is a lack of high-resolution imaging tools to study wound healing and scar formation systematically and longitudinally in vivo. Given collagen plays a critical role in the pathogenesis of abnormal wound healing, it is an important target for imaging. Increased collagen synthesis by fibroblasts with the expression of TGF-β and PDGF, and decreased collagen degradation with reduced MMP activity, are hallmarks of this process (Wolfram et al., 2009). In fact, a study demonstrated the use of TGF- β1 and 2 neutralizing antibodies in rats to prevent scar formation in dermal wounds (Shah et al., 1992). Therefore, our label-free, high-resolution imaging technique can serve as a promising tool to gain significant insights into wound healing by monitoring both scar size (using DOP) and collagen remodeling (using LR) longitudinally. For example, healing of the scar increases the LR signal and decreases the DOP back to values observed in normal skin.
Indeed, most of our current therapeutic approaches for treating HTS rely on targeting or reversing the abnormal collagen deposited during wound healing. For example, intralesional injection of corticosteroids is thought to decrease fibroblast proliferation and collagen synthesis, while pulsed dye laser treatment is thought to favor collagen degradation and fibroblast apoptosis (Urioste et al., 1999; Kuo et al., 2005). However, the lack of non-invasive scar imaging techniques in well-established animal models for HTS has hindered the study of HTS formation in vivo, especially at earlier time points since most studies relied on surgical scar specimens harvested months after their formation. Therefore, the ability to assess collagen in vivo in an animal model of HTS, as demonstrated in this paper, enables further investigation to explore novel, targeted therapeutic approaches for HTS, which is the subject of our ongoing studies. In addition, the PS-OFDI technique can be directly applied to imaging human HTS in vivo as we expect a similar optical signature (decreased local retardation and increased DOP) given the similarity in collagen organization, namely the presence of fine collagen fibers with haphazard orientation in a dermal nodule. Clinically, this non-invasive, high-resolution 3D scar imaging capability can be exploited to improve the management of HTS, including the initial assessment and longitudinal monitoring of treatment response in patients.
6-week old female Sprague-Dawley rats (~200g, N=18) were purchased from Charles River Laboratories (Wilmington, MA). The animals were housed in individual cages with access to food and water ad libitum, and were maintained on a 12-hour light/dark cycle in a temperature-controlled room. All animal procedures were approved by the Subcommittee on Research Animal Care (IACUC) of the Massachusetts General Hospital (protocol number 2012N000077) and were in accordance with the guidelines of the National Institutes of Health (NIH).
A biomechanical loading device was constructed from 22-mm expansion screws and metal supports, which allows the device to be placed over a healing incision without contacting the wound. The procedure was developed based on a previous report from Aarabi et al (Aarabi et al., 2007). After the induction of anesthesia with isoflurane, two 2 cm linear full-thickness incisions were made on the dorsum (5 cm apart; one at the cranial end and another at the caudal end) and then reapproximated with three 4-0 vicryl sutures. On post-incision day 4, the sutures were removed from the healing incisions, and the loading device was carefully secured with 6-0 nylon sutures on one of the incisions. The second incision served as a control. Tension on the wound was created by translating the expansion screws by 2 mm at the time of securing the loading device, and 2 mm every day thereafter. This was necessary to prevent stress relaxation due to the natural elongation of the skin resulting in a decrease in the tension on wounds. At the end of the loading period (4 days, 6 days, 8 days or 10 days), the device was removed and the scar was imaged 1 month later (or at 1-week intervals for the longitudinal study).
A 2.5cm × 2cm full-thickness excisional wound was created on the mid dorsum. The lax skin flanking the wound was gathered with a running 4-0 vicryl suture to create tension. On each side of the wound, excess skin was gathered in a running fashion starting at the cephelic aspect, down to the caudal aspect, and back up to the cephalic end. The length of the gathered skin was the length of the wound created (2.5cm). The small ridge of skin produced was tacked down laterally. This was performed on each side of the wound to secure the tension. A small piece of gauze and Tegaderm were placed over the wound. The sutures were left until reabsorbed (~2 weeks after excision). 0.05 mg/kg of buprenorphine was administered subcutaneously every 12h up to 72h after excision.
The PS-OFDI system used in this study has been reported in detail previously (Villiger et al., 2013). A wavelength-swept laser source operating at a center wavelength of 1320nm with an axial resolution of 9.4µm was used. The A-line rate was 54kHz and the signal was digitized at 85MHz. In the reference arm, an acousto-optic modulator was used to remove depth-degeneracy (Yun et al., 2004). An imaging window of 10 × 5 mm, consisting of 2048 A-lines/image × 256 images, was scanned with a focused beam featuring a lateral resolution of 15µm. The skin was apposed against a glass slide with ultrasound-gel as immersion liquid to center the superficial layers in focus. The polarization state of the input light was modulated between circular and linear polarization between adjacent A-lines, and the signal was detected with a polarization diverse receiver.
Reconstruction of the PS data was performed with spectral binning (Villiger et al., 2013), using 1/5th of the original spectral bandwidth (i.e. 9 overlapping spectral bins), a lateral Gaussian filter of a full width at half maximum (FWHM) equal to 12 adjacent A-lines, and an axial offset of 48µm to derive the local retardation. The DOP was evaluated independently for each spectral bin and input polarization state over the same lateral Gaussian kernel, and then averaged:
where Q,U,V and I are the spatially averaged, but un-normalized components of the Stokes vector, p denotes the input polarization states and n the spectral bin.
Research reported in this publication was funded in part by the National Institutes of Health P41 EB015903 grant and by the Shriners Foundation grant # 85120-BOS. W. C. Y. Lo was supported by the Canadian Institutes of Health Research (CIHR) Doctoral Foreign Study Award and M. Villiger was supported by the Swiss National Science Foundation.
CONFLICT OF INTEREST
The authors state no conflict of interest.