Assessment of tissue fibrosis in skin biopsies from patients with systemic sclerosis employing confocal laser scanning microscopy: an objective outcome measure for clinical trials?

Abstract

Objectives. To obtain an objective, unbiased assessment of skin fibrosis in patients with SSc for use in clinical trials of SSc disease-modifying therapeutics.

Methods. Skin biopsies from the dorsal forearm of six patients with diffuse SSc and six healthy controls, and skin biopsies from the forearm of one patient with diffuse SSc before and following 1 year treatment with mycophenolate mofetil were analysed by confocal laser scanning microscopy (CLSM) with specific antibodies against collagen types I and III or fibronectin. The integrated density of fluorescence (IDF) was calculated employing National Institutes of Health-ImageJ software in at least four different fields per biopsy spanning the full dermal thickness.

Results. The intensities of collagen types I and III and fibronectin IDF were 174, 147 and 139% higher in SSc skin than in normal skin, respectively. All differences were statistically significant. The sum of the IDF values obtained for the three proteins yielded a comprehensive fibrosis score. The average fibrosis score for the six SSc samples was 28.3 × 10⁶ compared with 18.6 × 10⁶ for the six normal skin samples (P < 0.0001). Comparison of skin biopsies obtained from the same SSc patient before treatment and after 12 months of treatment with mycophenolate mofetil showed a reduction of 39% in total fibrosis score after treatment.

Conclusions. CLSM followed by quantitative image analysis provides an objective and unbiased assessment of skin fibrosis in SSc and could be a useful end-point for clinical trials with disease-modifying agents to monitor the response or progression of the disease.

Introduction

SSc is a heterogeneous autoimmune disorder characterized by severe and usually progressive fibrosis in skin and multiple internal organs, microvascular dysfunction and humoral and cellular immune dysregulation [1–4]. The extent and rate of progression of skin fibrosis is critically important in determining the clinical features and prognosis of SSc and correlates with survival [5], functional limitations [6] and overall disease severity [7], therefore, skin involvement is the most widely used outcome measurement in SSc interventional trials [8]. Currently, there is no objective and quantitative measurement of skin fibrosis in SSc and the gold standard outcome measurement in clinical trials is the modified Rodnan skin score (mRSS) [9, 10]. Although the mRSS is accessible, non-invasive and cost-effective, it is a subjective assessment with substantial inter- and intra-observer variability calculated to be 25 and 12%, respectively [11]. Furthermore, it is not possible to differentiate fibrotic skin thickening from that resulting from oedema, inflammation or vascular engorgement. To reduce subjectivity and increase reliability, investigators have employed durometry and ultrasonography [12, 13]; however, these procedures display a similar inter-observer variability and a non-homogeneous sensitivity in uneven body regions.

Increased expression of the genes encoding interstitial collagen types I and III (COL I and -III, respectively) and of the production of the corresponding proteins are the hallmark of SSc and is responsible for the functional abnormalities in skin and various internal organs. Numerous studies have investigated circulating or urinary levels of collagen fragments as SSc biomarkers that may reflect the activity of the ongoing fibrotic process [14–17]. Another potential biomarker is COMP, a cartilage-specific protein recently also shown to be an important fibroblast product [18]. Significant correlations between serum COMP and the extent of SSc skin involvement and severity [19] and increased expression of COMP in SSc skin and cultured fibroblasts [20, 21] have been described. Histopathological quantitation of myofibroblasts has also been suggested as another potential biomarker of SSc skin fibrosis [22]. However, more extensive studies are required to validate these measurements as markers of tissue fibrosis in SSc.

The availability of an objective, reliable and reproducible quantitative method to assess the amount of skin fibrosis in SSc would allow for smaller sample sizes and enhanced detection of effective therapies in clinical trials. Here, we identified and quantified histopathological markers of fibrotic activity in SSc skin employing quantitative confocal laser scanning microscopy (CLSM) that can be used as reliable biological end-points for clinical trials of SSc disease-modifying therapeutic agents.

Patients and methods

Skin biopsies

Skin biopsies were obtained from the dorsal forearm of six untreated patients with rapidly progressive SSc of recent onset (>18 months from clinically detectable skin induration) fulfilling SSc classification criteria [23] with diffuse cutaneous involvement [24] following the Thomas Jefferson University Institutional Review Board-approved protocol. Normal skin samples were obtained from non-SSc patients undergoing unrelated surgical procedures. To examine the effects of a therapeutic intervention, skin biopsies were obtained from one patient with diffuse SSc of recent onset before and following 12 months’ treatment with 2.0 g/day mycophenolate mofetil (MMF) orally. The second biopsy was obtained from an area in close topographic proximity to the initial biopsy. The samples were fixed with 10% buffered formalin and embedded in paraffin.

Skin IF

After deparaffinizing the slides, tissue sections were incubated with a 1: 100 dilution of specific primary rabbit polyclonal antibodies (Rockland Immunochemicals, Gilbertsville, PA, USA) against COL I and -III, fibronectin (FBN) or α-SMA for 1 h at room temperature in a moist, dark chamber. Sections were washed twice with phosphate-buffered saline (PBS). To minimize non-specific staining, slides were incubated with fragment antigen binding prime (Fab′)-sheep anti-rabbit Cy3 antibody (Sigma, St Louis, MO, USA) for 1 h at room temperature, washed in PBS and then counterstained with DAPI-1500. No antigen retrieval was performed since commonly used antigen-retrieval procedures may cause variable extraction of tissue collagens or cell membrane alterations and cellular disruption resulting in increased background staining.

Analysis of fluorescence

Four different low power magnification (×20) fields consecutively spanning the full thickness of the dermis were analysed for each biopsy employing a Zeiss LSM 510 META CLSM (Zeiss, Wetzlar, Germany). The analyses were carried out in a blinded manner to disease/non-disease source of sample. A computer-generated 2.5D image that plots the intensity of fluorescence of a specific microscopic field was analysed with Image J software (NIH, Bethesda, Maryland, USA), which calculated the sum of the intensity of each pixel in a given microscopic field as the integrated density of fluorescence (IDF). The IDF value measures the overall fluorescence intensity in a given microscopic field. The average of the four IDF values per specimen was calculated. Statistical analysis of the average IDF values was performed using a two-tailed unpaired t-test. P < 0.05 was considered statistically significant.

Results

Six patients with diffuse SSc [four women, two men; median (s.d.) disease duration 8 (5) months, age 39 (11) years] were studied. The SSc biopsies were obtained from patients with early disease (<6 months for four out of six patients), the target population most commonly included in SSc disease-modifying clinical trials. Skin samples from six non-SSc subjects [four women, two men; age 43 (17) years] were used as normal controls.

Figure 1 shows the analysis of fluorescence intensity in normal and SSc skin samples. The left panels show representative images of the visual intensity of fluorescence, the centre panels show the computer-generated 2.5D analyses of the fluorescent images and the right panels show the Image J software-generated IDF values in arbitrary units. The Image J software-generated IDF value for COL I (Fig. 1A) shows a high correlation with the visual intensity of fluorescence and, most importantly, it is an accurate assessment of total COL I expression in the tissue section. The average COL I IDF in the normal skin was 5.82 (0.87) × 10⁶ vs 10.13 (0.92) × 10⁶ in the SSc skin (P = 0.0018). Figure 1B and C shows similar analyses for COL III and FBN. The average COL III in normal skin was 5.53 (0.69) × 10⁶ vs 8.10 (1.57) × 10⁶ in the SSc skin (P = 0.0197). The average FBN in the normal skin was 7.28 (0.51) × 10⁶ vs 10.10 (0.94) × 10⁶ in the SSc skin (P = 0.0086). The calculated amounts of COL I, -III and FBN were 174, 147 and 139% higher in SSc skin than in normal skin, respectively. All these differences were highly statistically significant. Analysis of α-SMA was of little value as the calculated IDF included, in addition to myofibroblast or activated fibroblast signals, the intense signal from smooth muscle cells in small arterioles or surrounding hair follicles.

Fig. 1.

Analysis of the fluorescence intensity for COL I (A), COL III (B) and FBN (C) in normal skin (N) and SSc skin (SSc). The left panels show the IF images, the centre panels show the computer-generated 2.5D image analysis plots of the corresponding microscopic fields and the right panels show the IDF of the two microscopic fields calculated by Image J software.

The average IDF values of the four microscopic fields from the dermis of each of the six SSc skin samples and six normal skin samples stained for COL I, -III and FBN were analysed by scatter plot (Fig. 2A). To obtain a comprehensive value of the amount of COL I, -III and FBN, we calculated and plotted the sum of the IDF values for all three extracellular matrix (ECM) proteins as the total fibrosis score for each of the six SSc skin samples and six normal skin samples (Fig. 2B). The average total IDF score was 28.3 × 10⁶ for the SSc skin samples compared with 18.6 × 10⁶ for the normal skin samples. A two-tailed, unpaired t-test indicated that the differences in expression were all statistically significant (P < 0.0001).

Fig. 2.

Comparison of IDF values between normal and SSc skin. (A) The average IDF for COL I, -III and FBN from four different microscopic fields for each of the six SSc skin samples and six normal skin samples (N) analysed by scatter plot. *P < 0.01; **P < 0.001. (B) The sum of the IDF values for COL I, -III and FBN of each of the six SSc skin samples and six normal skin samples was analysed as the total fibrosis score by scatter plot.The red dotted line demonstrates that there is no overlap between the total fibrosis scores calculated for SSc skin and normal skin samples. **P < 0.001.

To examine whether the fibrosis score is sensitive to change, we studied skin biopsies from one patient with diffuse SSc before and following 12 months of treatment with MMF. The second biopsy was taken in close topographic proximity to the initial biopsy site. The ECM protein content and fibrosis score were obtained following staining for COL I, -III and FBN. Two representative CLSM fields of the COL I staining in the dermis are shown in Fig. 3A and the plotted IDF values of seven separate microscopic fields spanning the dermis are shown in Fig. 3B. A profound and highly significant decrease in fluorescent intensity is observed in the sample following treatment. The sum of the average IDF values for COL I, -III and FBN of the SSc skin biopsies (IDF fibrosis score) showed a 39% reduction in the total fibrosis score following 1 year of oral MMF treatment (Fig. 3C).

Fig. 3.

(A) CLSM images of two representative fields of the dermis stained for COL I of a patient with diffuse SSc before treatment and after receiving oral treatment with 2.0 g/day MMF for 1 year. (B) The IDF values of seven microscopic fields stained for COL I of the same SSc patient before and after treatment were plotted. ***P < 0.001. (C) The sum of the average IDF values for COL I, -III and FBN of the SSc skin biopsy taken before and after treatment was analysed as the total IDF score. There is a reduction of 39% in fibrosis score after treatment with MMF for 1 year.

Discussion

The goal of this study was to obtain an assessment of skin fibrosis in SSc patients, which can be used as an objective and unbiased outcome measurement in clinical trials for SSc disease-modifying therapeutics. We determined the amounts of three ECM proteins known to be up-regulated in SSc skin employing CLSM in skin biopsies from six patients with diffuse SSc and six healthy controls. The amounts of COL I, -III and FBN were 174, 147 and 139% higher in SSc skin than in normal skin, respectively. All differences were statistically significant. In order to obtain an overall assessment of the amount of dermal fibrosis, the sum of the IDF values for all three ECM components was calculated as a total fibrosis score for each of the samples. The total fibrosis score demonstrated a more pronounced difference between SSc and normal skin compared with the individual IDF values obtained separately for each ECM protein. This observation may be explained by differences in the stage of the disease that may result in increased abundance of a particular ECM protein at certain stages of disease progression, and therefore a global or total fibrosis score may be more representative of the overall level of fibrosis in the tissue.

Our results also showed that the measurements were sensitive to change over time, as they displayed a highly statistically significant reduction (39%) following 12 months of oral treatment with MMF, a drug that has been recently suggested as a potentially effective anti-fibrotic agent [25, 26] and has been used as a disease-modifying therapy for SSc-associated lung disease [27–30] and dcSSc of recent onset [31, 32]. However, it was not possible to demonstrate a direct correlation between the fibrosis score values obtained with this procedure and the mRSS or the durometer assessment of skin induration in a previous study [12] (data not shown). The failure to establish these correlations suggests that the parameters evaluated by mRSS and durometer measurements are not the same as those being evaluated by the procedure described here: CLSM provides a direct measurement of ECM accumulation, whereas mRSS and durometer measurements reflect other parameters including tissue oedema, inflammation, tissue tethering, vascular engorgement, etc.

The assessment described here is a substantial improvement compared with the qualitative and highly subjective mRSS assessment, since the values are obtained with an objective and unbiased (‘investigator interpretation free’) method and the assessment generates absolute numbers that can be used as an objective outcome measurement of tissue fibrosis in SSc. The values obtained by analysing and quantifying the IF intensity are sensitive enough to clearly distinguish between fibrotic and non-fibrotic skin. However, a potential problem that may need to be evaluated in additional studies is the possible masking of antigenic determinants when different samples are compared owing to altered and different composition of the ECM in diseased or non-affected tissues that could affect the penetration of the antibodies.

We do not foresee that CLSM followed by quantitative image analysis would be widely applicable in the clinical practice setting since skin biopsies are not typically necessary to establish the diagnosis. However, this objective method may be used in clinical trials to compare the amount of tissue fibrosis before and following administration of a disease-modifying drug as part of a clinical trial and to accurately monitor the response or progression of the disease. Further assessment of a larger number of patients will be required to validate this procedure.