A Study of Collagen Refractility in Dermatofibroma and Dermatofibrosarcoma Protuberans using Diffractive Microscopy

Abstract

Background:

Diffractive microscopy creates contrast within samples that are otherwise uniform under bright light. This technique can highlight subtle differences in refractive indices within birefringent samples containing varying amounts of mature collagen. Dermatofibroma (DF) and dermatofibrosarcoma protuberans (DFSP) possess differences in their mature collagen content, and therefore may be distinguishable using diffractive microscopy.

Methods:

Two hundred forty-two DF and 85 DFSP hematoxylin/eosin-stained specimens were analyzed using diffractive microscopy. Data regarding the distribution pattern and strength of refractility was recorded.

Results:

DFSP was more frequently found to be focally, weakly, or non-refractile (82.9%; n=68) under diffractive microscopy, while DF more often showed diffusely bright refractility (52.9%; n=128). DFSP samples with diffuse refractility in portions of the lesion (17.1%, n=14) also exhibited a unique checkerboard pattern distinct from that which was seen in DF samples.

Conclusions:

The absence of diffuse refractility was more closely associated with DFSP, as was the presence of a unique checkerboard diffraction pattern. Despite high sensitivity (Sn=82.9%), absent refractility was not a specific test (Sp=52.9%), with 47.1% (n=114) DF samples sharing this feature. The distinction between DF and DFSP is often diagnosed using H&E alone. In difficult cases, examination of collagen under diffractive microscopy may be useful in distinguishing DFSP from DF and provide an alternative cost-effective tool to immunohistochemical staining.

Introduction

Dermatofibroma (DFs) are commonly confined to the dermis and superficial subcutaneous tissue while dermatofibrosarcoma protuberans (DFSP) has a propensity to invade deep tissue structures. Immunohistochemical techniques are not entirely sensitive nor specific; between 6-10% of DFSP are CD34-negative and 15-25% are Factor XIIIa positive.^1–3 Identification of translocation t(17;22)(q22;q13), which results in COL1A1-PDGFB fusion in DFSP, may be helpful in difficult cases, but each of these techniques add time and expense.^2,4

Collagen within the reticular dermis is arranged in an organized yet non-uniform meshwork of cross-linked fibrils. This precise anisotropic configuration provides the strength and elasticity necessary for the skin to accommodate external forces.⁵ Anisotropic materials contain two or more refractive indices which are sensitive to orientation. As incident light is enters the sample, it is refracted into two polarized rays oriented at right angles to one another.^5,6 The resultant effect on light is termed birefringence. In bright light microscopy, there is often not sufficient contrast for the human eye to detect these differences in phase or refractive index, even with colored staining. The principle of diffraction, however, has been utilized to create contrast within samples containing small differences in refractive index that are otherwise indistinguishable.

Diffraction describes the process by which light is bent as it passes by an edge or through a small aperture. Diffracted light produces first order side bands which are slowed down by one quarter-wavelength out of phase from the undiffracted or “direct” light passing through the sample. When light is recombined at the microscope’s objective, the diffracted edges, outlines and refractive index gradients appear more intensely illuminated.^7,8

Coherent and oblique illumination, phase contrast, differential interference contrast (DIC), confocal microscopy and dark-field techniques all utilize diffraction to visualize features with low contrast. Using a standard light microscope, various settings can be manipulated to create diffraction as well.^7,8 Both reducing the size of the condenser aperture diaphragm and lowering the condenser, for example, increase contrast and depth of field at the expense of reduced resolution.⁹

Manual “hand” diffraction is an alternative technique in which the microscopist gradually moves uses his or her hand horizontally to interfere with the light column, which causes bending of light around the obstruction.¹⁰ This enhances diffraction at the junction between bright and dark portions of the slide. This diffraction technique has been shown to enhance refractility in pigmentary conditions including minocycline and imipramine discoloration, argyria, and in samples containing silica and hemosiderin deposits. This has also been used to generate contrast between bright dermal collagen and dark, nonrefractile areas of scar and tumor stroma within samples of basal cell carcinoma, desmoplastic nevi, alopecia, and lichen planopilaris.¹⁰

This study aimed to analyze DF and DFSP samples using this manual diffraction technique. While DF and DFSP are often diagnosed using hematoxylin/eosin (H&E) microscopy alone, difficult cases may require additional use of immunohistochemical (IHC) staining, molecular studies or cytogenetics. It was hypothesized that DF and DFSP would exhibit differences in refractility and that diffractive microscopy could provide an addition tool in their diagnosis.

Methods and Materials

A retrospective study was performed using archived H&E-stained histologic slides. Three hundred thirty-one histologic slides were analyzed. Four DF and three DFSP samples were excluded due to poor slide quality. A total of 242 DF and 85 DFSP specimens were included in final data collection. Four out of 82 DFSP specimens (4.9%) and 13 out 242 DF specimens (5.4%) contained variant tumor subtypes (Table 2).

Table 2.

DF and DFSP Subtypes

Diagnosis	Subtypes	Number of Cases, n
DF (n=246)	DF unspecified	229
	Keloidal	3
	Cellular	3
	Aneurysmal	2
	With monster cells	2
	Glandular	1
	Diffuse linear	1
	With focal osteoid change	1
	Excluded (poor slide quality)	4
DFSP (n=85)	DFSP unspecified.	78
	Bednar	3
	Myxoid	1
	Excluded (poor slide quality)	3

Slides were analyzed under light microscopy using the manual “hand” diffraction technique described above and depicted in Figure 1. Categorical data was recorded regarding the distribution pattern and strength of refractility under diffraction microscopy, as described in Table 1. Specimens in which the majority (>50%) of tumor demonstrated diffusely bright refractile collagen with were recorded as “bright diffraction present.” Specimens in which either only focal areas of tumor were highly refractile, refractility was weak, or the specimen was entirely nonrefractile were recorded as “bright diffraction absent.” All microscopic analyses were conducted by a single physician reviewer within a one month time-frame in order to maintain internal consistency.

Figure 1.

Manual “hand” diffraction involves the microscopist moving his or her hand horizontally to partially obstruct the incident light column.

Table 1.

Diffractive Microscopy Grading Criteria

Feature	Categorical Grading	Criterion definition
Bright Diffraction	Present	Diffuse refractility comprising >50% of tumor, with increased brightness compared to non-tumor cells
	Absent	Focal refractility, weak refractility, or entirely absent of refractility within the bulk of the tumor (>50%)

A statistical program IBM^® SPSS^® Version 29 was used for data processing. Chi-squared tests were used to compare frequencies of collagen diffraction between the DFSP and DF samples. Sensitivity, specificity, positive and negative likelihood ratios [LR (+) and LR (−)] were calculated to study the diagnostic relevance of collagen diffraction. In this analysis, DFSP was considered the “disease state.” Given our hypothesis that DFSP samples exhibited a lesser degree of refractility under diffractive microscopy, the absence of bright diffraction was considered a “positive test.” LRs can be used to multiply a pre-test probability of disease (i.e. DFSP in this instance) to obtain a post-test probability. A LR (+) of more than 10 would indicate that the absence of bright diffraction increases the probability that the sample is DFSP, while LR (+) of 5-10 indicates a moderate probability, and less than 5 indicates a weak association with DFSP. A LR (−) of less than 0.1, on the other hand, would indicate that the presence of bright diffraction increases the probability the sample is a DF, 0.1 – 0.5 would indicate a moderate probability, and more than 0.5, a weak probability with DF.¹¹

Results

When used to examine DF and DFSP samples, manual diffraction yielded a significant difference in refractive patterns between the two tumor types (p<0.001) as shown in Table 3. DFSP was more frequently found to be focally, weakly, or non-refractile under diffractive microscopy (82.9%; n=68) (Figure 3). However, 47.1% (n=114) DF samples also demonstrated “absent” diffraction. DF was more often found to be diffusely and intensely refractile (52.9%; n=128) (Figure 2), with only 17.1% (n=14) of DFSP samples exhibiting similar diffuse refractility. DFSP with diffuse refractility often exhibited a bright checkerboard pattern unique from the diffuse diffraction seen in DF samples (Figure 3B). DF without diffraction was predominant histiocytic with little mature collagen.

Table 3.

Collagen Diffraction of DF and DFSP

Collagen Diffraction	DF N (%)	DFSP Nn (%)	Differences in frequency of features, p-value	Sensitivity	Specificity	LR (+)	LR (−)
Absent (Nonrefractile, focal, weak)	114 (47.1%)	68 (82.9%)	<0.001	82.9%	52.9%	1.76	0.32
Present (Diffuse, brilliant)	128 (52.9%)	14 (17.1%)

Figure 3.

(A) DFSP exhibiting absence of bright diffraction (central) compared to bright diffraction of surrounding normal collagen (*) at x20 magnification. (B) DFSP exhibiting characteristic diffuse checkerboard diffraction at x40 magnification.

Figure 2.

(A) Dermatofibroma (DF) demonstrating diffraction at x40 magnification. (B) DF demonstrating diffraction at x40 magnification. (C) DF demonstrating diffuse diffraction at x40 magnification. (D) DF demonstrating diffuse diffraction x20 magnification.

Sensitivity, specificity, LR (+) and LR (−) calculations are presented in Table 3 and are based on presence or absence of refractility on a bimodal scale. The absence of refractility under diffractive microscopy was found to be a relatively sensitive test (Sensitivity 82.9%) but less specific (Specificity 52.9%) for DFSP. Given a positive test result (in this case, the absence of bright diffraction), the likelihood a specimen is DFSP, rather than DF, is increased 1.76-fold (LR (+) 1.76). In contrast, the presence of diffuse refractility reduces the probability of a DFSP diagnosis between 0 and 15% (LR (−) 0.9). When attention is paid to the pattern of refractility (diffuse vs checkerboard), this simple test becomes even more useful. Positive and negative predictive value were not calculated, as the samples used in this study were obtained from an archived teaching library, and therefore disease frequencies were not necessarily representative of prevalence in clinical populations.

Discussion

Birefringence has been studied in numerous collagen subtypes under polarized microscopy within human and animal models. These studies have found differences in collagen birefringence due to photoaging, thermal damage, and throughout keloid and hypertrophic scar maturation.^{6, 12–15}

Due to differences in mature collagen content, DF and DFSP demonstrated appreciable differences using a simplified manual diffraction technique, Our findings suggest the absence of refractility under manual diffraction has a relatively high sensitivity (Sn=82.9%) for DFSP, but is not specific, with 47.1% of DF samples also demonstrating absent refractility in portions of the lesion. The presence of diffuse refractility within a sample indicates a low likelihood of DFSP. Additionally, DFSP that demonstrates bright diffraction does so in a characteristic checkboard pattern that is easily recognized (Figure 3B).

Limitations of manual diffraction as a diagnostic tool include subjective interpretation of results. Light microscopes are, by nature, diffraction-limited, and a trade-off exists between obtaining sufficient contrast while both limiting diffraction artifact and achieving adequate resolution.¹⁸ This limitation is minimized by examining samples at low-magnification. Limitations of this study design include bias introduced by the use of archived teaching slides for analysis. Our findings suggest that collagen refractility may be useful in distinguishing DFSP from DF, in that an absence of diffuse refractility is more closely associated with DFSP. When diffraction is present in DFSP, it tends to demonstrate a characteristic checkerboard pattern (Figure 3B). Manual diffraction may be a helpful tool in diagnosis and a offer an alternative to IHC techniques.