Diagnosis and differential diagnosis of dermatofibrosarcoma protuberans: Utility of high‐resolution dynamic contrast‐enhanced (DCE) MRI

Abstract

Background

Dermatofibrosarcoma protuberans (DFSP) is a kind of low‐grade malignant spindle cell neoplasm, the diagnosis, and treatment, which have markedly attracted clinicians’ attention for its repeated recurrence. High‐resolution magnetic resonance imaging (HR‐MRI) has shown unique capabilities in diagnosis of various cutaneous tumors.

Materials and methods

Data of 29 patients with clinically suspected DFSPs and undergoing dynamic contrast‐enhanced (DCE) HR‐MRI preoperatively were prospectively collected. The HR‐MRI qualitative features were evaluated and compared. The DCE‐associated quantitative parameters and the time‐signal intensity curve (TIC) types were provided using DCE sequences.

Results

A total of 7 DFSPs, nine dermatofibromas (DF, including four cases of cellular variant [CDF]), 12 keloids, and one nodular fasciitis were enrolled. DFSP showed the largest major diameter and the deepest depth. Five DFSPs (71.4%) showed ill‐defined margins as well as infiltration of peripheral adipose. All DFSPs showed irregular shape. Most DFSPs presented hyperintensity on T₂WI (71.4%) and iso‐intensity on T₁WI (85.7%). Six cases (85.7%) had significant enhancement, and six cases (85.7%) had homogeneous enhancement. There were significant differences of K ^trans, K_ep , V_e and iAUC values among DFSPs, DFs, and keloids, and DFSP had the highest values for these parameters. Six DFSPs (85.7%) and four CDFs (100%) showed type‐III TICs, while the other lesions showed type‐Ⅰor type‐Ⅱ TICs.

Conclusions

DCE‐HR‐MRI could show the growth characteristics of DFSPs, which was of great value for the diagnosis and differential diagnosis of DFSPs and was helpful for the determination of treatment options, thereby to improve the prognosis of patients.

1. INTRODUCTION

Dermatofibrosarcoma protuberans (DFSP) is an infrequent cutaneous fibrohistiocytic, slow‐growing, and low‐to‐intermediate‐grade malignant neoplasm, accounting for 1% of all soft tissue sarcomas, with an incidence of 0.8–5 cases per million people each year. ¹ DFSP mainly presents as a painless cutaneous lesion that may appear as a violet or pink plaque located at the trunk or at the proximal extremities. ² DFSP is easily misdiagnosed as several benign tumors, mainly including dermatofibroma (DF, also known as fibrous histiocytoma), keloids, and nodular fasciitis (NF), ³ especially for the cellular DF variant (CDF). ⁴ These lesions have similar manifestation with DFSP, further, owing to the same fibroblastic or myofibroblastic origin as DFSP, and histomorphology and immunophenotype show a broad range of overlap among these lesions. These lesions all entirely or predominantly composed of spindle cells microscopically. ⁵

However, the treatment and prognosis of these lesions are quite different. DFSP is a locally aggressive tumor, while it rarely leads to distant and local lymph node metastasis (< 5%). ⁶ DFSP has a high recurrence rate after standard excision (SE), and the recurrent lesions are more locally destructive, more prone to sarcomatous transformation and metastasis. ⁷ A previous study demonstrated that wide local excision (WLE) with resection margin of 2–3 centimeter (cm) resulted in a decreased recurrence rate. ³ Mohs micrographic surgery (MMS) is a highly appropriate surgical approach, ⁶ while it is complex, time‐consuming, and costly when the tumor invaded deeply. ¹ DF is one of the most common benign cutaneous tumors, with multiple variants, including cellular, aneurysmal, atypical, keloidal, and myxoid types. ⁵ , ⁸ , ⁹ CDF is characterized by abundant spindle cells with increased mitotic activity, which may also easily recur locally after incomplete resection, ⁴ , ⁸ especially lesions with a size of greater than 1 cm. Although as a kind of benign tumor, some researches have reported the distant metastasis of CDF. ¹⁰ , ¹¹ , ¹² , ¹³ WLE ¹⁰ and MMS ⁴ could also decrease the recurrence rate of CDF. To sum up, DFSP and CDF have a lot in common in recurrence, “metastasis” and resection. Keloid is a kind of reactive lesion, which is composed of unorganized proliferation of fibrous tissues, arising from a site of injury due to an aberrant healing process (it is noteworthy that approximately 10% of DFSP ¹⁴ and part of DF ¹⁵ are also related to prior skin trauma). A number of scholars pointed out that keloids are uniquely aggressive, accompanying with a progressive peripheral expansion behavior, as well as a certain recurrence rate after treatment, suggesting that keloids possess particular oncological features. ¹⁶ NF is a kind of benign, self‐limiting, reactive tumor, while it is easily misdiagnosed as a malignant lesion ¹⁷ , ¹⁸ because of its rapid proliferation and pseudosarcomatous morphological appearance. NF may gradually disappear without surgical intervention. ¹⁷

Some studies ¹⁹ , ²⁰ , ²¹ have shown the features of DFSP in computed tomography and magnetic resonance imaging (MRI). But few researches paid attention to the differential of DFSP. High‐resolution‐MRI (HR‐MRI) is a novel, noninvasive imaging technique, which can present cutaneous lesions as small as 1–2 mm with high resolution. It possesses multiplanar and multidirectional imaging capabilities. Compared with conventional skin imaging methods, with the ability of visualization of wholly morphological features, HR‐MRI can easily show details of tumors simultaneously, which indicates its high clinical significance for the diagnosis of skin tumors. ²² Dynamic contrast‐enhanced (DCE)‐MRI can reflect perfusion and vascular permeability of living tissues based on the dynamic changes in signal intensity after contrast agent injection. ²³ As a functional imaging method, DCE‐MRI has been widely used in the diagnosis of diverse types of cancer, such as breast cancer, ²⁴ prostate cancer, ²⁵ ovarian cancer, etc. And our previous research had revealed the utility of DCE‐HR‐MRI in the diagnosis of cutaneous squamous cell carcinoma. ²²

At present, preoperative diagnosis of DFSP mainly depends on histopathology and immunohistochemistry (IHC) examinations following needle biopsy. However, the diagnosis is occasionally associated with uncertainties. ⁵ As a noninvasive and objective imaging method for cutaneous lesions, in the present research, we aimed to assess the clinical utility of DCE‐HR‐MRI in the diagnosis and differential diagnosis of DFSP.

2. MATERIALS AND METHODS

2.1. Subjects

The inclusion criteria are as follows: (1) Patients who were clinically suspected with DFSPs and underwent DCE‐HR‐MRI, (2) Patients who underwent skin tumor excision within 1 month after DCE‐HR‐MRI, and (3) Patients who obtained exact histopathological results. The exclusion criteria are as follows: (1) Patients with common MRI contraindications and (2) Patients with poor‐quality MR images. Patients who were admitted to our hospital from June 2019 to August 2021 and met the above‐mentioned criteria were prospectively enrolled in this study. This study has been approved by the institutional review board, and written informed consents were obtained from all patients.

2.2. HR‐MRI examination

All DCE‐HR‐MRI examinations were performed on a 3.0 Tesla (T) MRI system (Magnetom Verio, Siemens Healthcare, Erlangen, Germany), using a dedicated 7‐centimeter‐diameter single‐element surface coil with 32 receiver channels and serial number of gradient system‐engine gradients (maximum gradient field strength, 45 mT/m; slew rate, 200 T/m/s). The small loop radiofrequency receiver coil was placed as close to the clinical area of interest as possible, preferably flushing with the skin, without causing distortion of the lesion or underlying skin. The tumor was centered in the middle of the coil. In cases where the receiver coil could not lie flush with the skin surface, such as supraclavicular fossa lesions, the coil was placed carefully as central as possible. Tapes and sandbags were used to secure the coil to the area of interest, or to prevent motion artifacts in necessary conditions. Two routinely acquired MRI protocols, including T₁‐weighted imaging (T₁WI) and T₂‐weighted imaging (T₂WI), were utilized. Parameters of the sequences are as follows: T₁WI (repetition time [TR], 612 ms; echo time [TE], 14 ms; signal averaging, one time; field of view [FOV], 80 × 80 square millisecond [mm²]; matrix size, 320 × 320; section thickness, 1–5 mm; turbo factor, 3; voxel size, 0.3 × 0.3 × 2.0 mm; acquisition time, 93 s [s]); T₂WI (TR, 4790 ms; TE, 96 ms; signal averaging, two times; FOV, 80 × 80 mm²; matrix size, 384 × 384; section thickness, 1–5 mm; turbo factor, 6; voxel size, 0.2 × 0.2 × 2.0 mm³; acquisition time, 169 s). T₁WI commonly scans axial planes and alternative sagittal or coronal planes depending on lesion location and shape, to obtain the three‐dimensional morphology. Bi‐flip‐angle T₁ mapping was herein carried out prior to DCE sequence, with flip angles of 2° and 15°. The parameters of the two T₁ mapping sequences are as follows: TR, 5.14 ms; TE, 1.39 ms; FOV, 142 × 100 mm²; matrix size, 256 × 256; slice thickness, 3.0 mm. DCE sequence was undertaken after intravenous administration of 0.2 mmol/kg gadopentetate dimeglumine (Gd‐diethylenetriaminepentaacetic acid [DTPA]) at a rate of 3 ml/s. The DCE sequence parameters are as follows: TR, 4.21 ms; TE, 1.42 ms; signal averaging, one time; FOV, 142 × 142 mm²; matrix size, 256 × 256; section thickness, 3 mm; voxel size, 0.6 × 0.6 × 3.0 mm³; and acquisition time, 306 s. After injection of Gd‐DTPA, enhanced T₁WI sequences with and without fat suppression could be achieved.

2.3. Evaluation of HR‐MRI findings

Two radiologists (RH and YL, with 5 and 15 years of experience in MRI diagnosis, respectively), who were blinded to all histopathological results and clinical information, evaluated qualitative HR‐MRI characteristics independently. Disagreements were resolved through discussions between them. Characteristics included major diameter, depth, signal intensity in T₁WI and T₂WI, morphology, margin, cutaneous involvement layer, enhancement degree, and enhancement pattern of lesions. The major diameter was defined as the longest diameter in any planes. The depth referred to the maximum length from the basal of the lesion to surface of skin. These two values were measured independently by the two radiologists, and their average was calculated. The signal intensity in T₁WI and T₂WI were described as iso‐intensity or hyper‐intensity compared with muscles. The morphologies were divided into round or oval, flat spindle, and irregular mass. The margins were described as well‐defined or poorly‐defined margins. The cutaneous involvement layers included dermis, subcutis (composed of subcutaneous fat mainly), deep fascia (a layer of dense connective tissue that surrounds muscles ²⁶ ), and muscular layer. The enhancement degrees were divided into mild‐to‐moderate and significant enhancement, according to the change of T₁WI intensity values ​​before and after Dd‐DTPA enhancement (when [postenhancement value–preenhancement value]/preenhancement value > 1, enhancement was regarded as significant). The patterns of enhancement were categorized into homogeneous and inhomogeneous types.

Another radiologist (QY, with 3 years of experience in MRI diagnosis), who was also blinded to histopathological and clinical information, drew oval regions of interest (ROIs) for each case using a commercial postprocessing software (Tissue 4D, Siemens Healthcare). DCE‐associated quantitative parameters of ROIs were automatically calculated on the basis of the modified hemodynamic two‐compartment Tofts model, including endothelial transfer constant (K ^trans), reflux rate (K_ep ), fractional extravascular extracellular space volume (V_e), and initial area under the curve (iAUC). To reduce bias as far as possible, 1–2 cm² ROIs were thrice placed, as large as possible, in the homogeneous enhancement areas, avoiding formation of hemorrhagic, cystic, necrotic areas, and vessels. If the lesion was large and heterogeneous, the ROIs that met the above‐mentioned criteria meanwhile with the maximum iAUC values (i.e., the highest degree of enhancement) were selected. Time‐signal intensity curves (TICs) were plotted using the MeanCurve software (Siemens Healthcare). We attempted to place the ROIs in an identical site to the ROIs generating quantitative parameters in each case. Three types of TICs are defined as follows: Type‐Ⅰ, a slow, gradual, constant enhancement within 5 min (min). Type‐Ⅱ, an early significant enhancement within 1–2 min and fixing at a high level continuously. Type‐III, an early steep and rapid enhancement, then descending subsequently.

2.4. Histopathological examination

All the eligible patients underwent surgical resection within 1 month after DCE‐HR‐MRI examination. Specimens were stained by hematoxylin and eosin. IHC including vimentin, CD34, KI ‐ 67, desmin, α‐smooth muscle actin (SMA), CD68, CD56, and so on was carried out. Two pathologists (LS and XZ, with 5 and 10 years of experience in dermatopathology, respectively) reviewed all available histopathology slides.

2.5. Statistical analysis

The continuous variables were presented as mean ± standard deviation, and the categorical parameters were shown as count (percentage). All statistical analysis was performed using SPSS 23.0 software (IBM, Armonk, NY, USA). The statistical significance was set to p < 0.05. Age of patients, major diameter, depth, and DCE‐associated quantitative parameters were compared using one‐way analysis of variance (ANOVA). Fisher's exact test or chi‐square test was used to compare differences of gender, part of qualitative features, and type of TICs. When it comes to comparison of TIC types, as mentioned above, DFSPs and CDFs were allocated into A group because of highly similarity in manifestation, treatment, and prognosis, while the other lesions were allocated into B group. The only case of NF was not involved in the analysis of quantitative and qualitative parameters (except for the TIC type).

3. RESULTS

3.1. Characteristics of study subjects

A total of 29 cases (38.7 ± 15.6 years, male/female = 15/14) were eventually enrolled, including seven cases of DFSP (42 ± 18.3 years, male/female = 5/2), nine cases of dermatofibroma (44.3 ± 20.4 years, male/female = 3/6), 12 cases of keloid (32.0 ± 6.9 years, male/female = 7/5), and one case of NF (44 years, female). There were no significant differences in age and gender among patients with DFSPs, DFs, and keloids (p = 0.163, 0.351, respectively). The tumors were located in extremities (n = 9), trunks (n = 18), and necks (n = 2). Besides, five cases grow subcutaneously but could touch, and 24 cases manifested as nodules or masses protruding from surface of skin, color ranging from pink to tan.

3.2. HR‐MRI characteristics

Of all DFSPs, three cases (3/7, 42.9%) underwent postoperative recurrence (one case underwent repeated recurrence for several times), and the recurrence interval was 1–8 years. DFSPs had the largest major diameters (34.4 ± 17.6 mm, p = 0.024) and depths (15.7 ± 10 mm, p = 0.04, Table 1). Majority of DFSPs showed hyperintensity in T₂WI (5/7, 71.4%) and iso‐intensity in T₁WI (6/7, 85.7%) (Figures 1 and 2), two of which showed mixed intensity due to bleeding and sarcomatous transformation (Figures 1). All DFSPs presented as irregular masses regardless of their size. Five cases (5/7, 71.4%) had poorly‐defined margins (infiltration of surrounding adipose could be observed). Specific features included tiny spinous protrusion, crab‐foot‐like spicules and satellite lesions (Figure 3)]. Six DFSPs (6/7, 85.7%) were detected with homogeneous enhancement pattern, while the other one had inhomogeneous enhancement pattern (Figure 1). Six of seven cases (85.7%) showed significant enhancement, and one case showed mild‐to‐moderate enhancement. HR‐MRI displayed that four DFSPs (4/7, 57.1%) involved deep fascia, and three (3/7, 42.9%) involved subcutaneous fat.

TABLE 1.

Population and HR‐MRI characteristics of DFSP and other confused lesions

HR‐MRI features	Dermatofibrosarcoma protuberans (n = 7)	Dermatofibroma (n = 9)	Keloid (n = 12)	p‐Value	Nodular fasciitis (n = 1)
Age (years) †	42.0 ± 18.3	44.3 ± 20.4	32.0 ± 6.9	0.163	44
Male/Female ‡	5/2	3/6	7/5	0.351	0/1
Major diameter (mm) †	34.4 ± 17.6	14.7 ± 12.4	30.0 ± 13.9	0.024	28.1
Depth of lesion (mm) †	15.7 ± 10.0	7.2 ± 5.3	4.8 ± 1.8	0.04	15.7
Location
Trunk	6 (85.7%)	2 (22.2%)	10 (83.3%)		0
Extremities	1 (14.3%)	6 (66.7%)	2 (16.7%)		0
Neck	0	1 (11.1%)	0		1
T1WI intensity
Hyperintensity	0	0	4 (33.3%)
Iso‐intensity	6 (85.7%)	8 (88.9%)	8 (66.7%)		1
Mixed intensity	1 (14.3%)	1 (11.1%)
T2WI signal
Hyperintensity	5 (71.4%)	0	0		1
Iso‐intensity	0	7 (77.8%)	12 (100%)
Mixed intensity	2 (28.6%)	2 (22.2%)	0
Morphology
Round or oval	0	8 (88.9%)	1 (8.3%)
Flat spindle	0	0	11 (91.7%)
Irregular mass	7 (100%)	1 (11.1%)	0		1
Margin ‡				0.002
Well‐defined	2 (28.6%)	6 (66.7%)	12 (100%)		1
Poorly‐defined	5 (71.4%)	3 (33.3%)	0
Involvement layer
Dermis	0	2 (22.2%)	10 (83.3%)
Subcutis	3 (42.9%)	6 (66.7%)	2 (16.7%)
Deep fascia	4 (57.1%)	0	0		1
Muscle	0	1 (11.1%)	0
Enhancement pattern ‡				0.097
Homogeneous	6 (85.7%)	6 (66.7%)	12 (100%)		1
Inhomogeneous	1 (14.3%)	3 (33.3%)	0
Enhancement degree ‡				0.008
Mild to moderate	1 (14.3%)	4 (44.4%)	10 (83.3%)
Significant	6 (85.7%)	5 (55.6%)	2 (16.7%)		1

Abbreviations: DFSP, dermatofibrosarcoma protuberans; HR‐MRI, high‐resolution magnetic resonance imaging.

^† Values are shown as mean ± standard deviation (SD). p‐Values ​​were calculated using one‐way ANOVA.

^‡ p‐Values were calculated using the chi‐square test or Fisher's exact test.

FIGURE 1.

A 66‐year‐old man with a fibrosarcomatous dermatofibrosarcoma protuberans (DFSP) at his right shoulder, with a size of about 5.3 × 4.8 × 3.6 cm³, who had undergone excision due to a DFSP in the same location within 8 years before his admission to our hospital. The patient complained of a new mass that grew rapidly during 2 years. (A) Macro‐examinations showed that the mass was translucent with a firm texture. (B–D) The mass was presented iso‐intensity in T₁WI (B) and inhomogeneous hyperintensity in T₂WI (C), and an annular with hypo‐intensity T₂ signal (capsule) could be observed. The tumor grew beyond the capsule. The mass appeared with a significant and inhomogeneous enhancement pattern in fat‐suppression (FS)‐T₁WI (D), while multiple tiny spinal protrusions infiltrating adjacent subcutaneous fat layer could be clearly visualized (arrows in D). (E) Time‐signal intensity curve (TIC) of the oval ROI in “D” was type‐III TIC. (F) Dense spindle cells were arranged in a fascicular pattern, and a frequent mitosis could be observed (hematoxylin‐eosin, H&E, x100). (G) The region in the square (D) showed degenerative sarcoma cells and abundant blood vessels. (H) The tumor cells infiltrated into the subcutaneous fat tissue in a honeycomb pattern (H&E, x100)

FIGURE 2.

A 25‐year‐old man with a dermatofibrosarcoma protuberans (DFSP) at his back for 3 years, with a range of about 5.7 × 4.2 × 1 cm³. (A) Visually normal in skin surface. (B–D) The lesion was shown with iso‐intensity in T₁WI (B) and hyper‐intensity in T₂WI (C), and basal of the lesion was adjacent to deep fascia, while a thin layer of fat could be observed. Enhanced fat saturation (FS)‐T₁WI (D) showed that the lesion had a poorly‐defined margin with subcutaneous fat. (E) Type‐I time‐signal intensity curve (TIC) of the oval regions of interest (ROIs) in D. (F) Diffuse, uniform spindle cells were arranged in a strip‐shaped pattern (H&E, x100). More homogenized red‐stained collagens could be observed. (G) The tumor cells infiltrated into subcutaneous fat tissue diffusely (hematoxylin and eosin [H&E], x40). (H) CD‐34 was positively diffusely expressed (x40)

FIGURE 3.

Specific features of dermatofibrosarcoma protuberans (DFSP) infiltrating surrounding fat. (A–C) Tiny spinous protrusions stretching along fatty space, lengthened up to 1.29 cm (A) coud be observed, indicating the necessity of wide local excision (WLE). (D) A DFSP case in the abdominal wall. Crab‐foot‐like spicules of the basal (short arrow), which grow to deep fascia and the satellite lesion (long arrow), were displayed

Regarding DF, there were four cases of CDF, four cases of conventional (also knowns as fibrous) type and one case of keloidal type. Two CDFs were detected with the first postoperative recurrence, and the recurrence intervals were 1 and 19 months, respectively. Seven DFs (7/9, 77.8%) presented iso‐intensity in T₂WI, while eight DFs (8/9, 88.9%) showed iso‐intensity in T₁WI. There were other one DF, which had mixed T₁WI and T₂WI intensity (Figure 4), and one DF having mixed intensity in T₂WI. Majority (8/9, 88.9%) DFs presented as round or oval shapes. Only one CDF had a lobulated morphology (Figure 4, two nodules enveloped by complete capsules could be observed within the lesion). Most CDFs (6/9, 66.7%) had well‐defined margins, while three CDFs showed vague margin (Figures 4 and 5). Six DFs (6/9, 66.7%) showed homogeneous enhancement pattern. Furthermore, four CDFs and one keloidal type had significant enhancement, while four cases of conventional type were mildly‐to‐moderately enhanced. HR‐MRI showed eight DFs (8/9, 88.9%) merely involved dermis and superficial subcutaneous fat, and one CDF (1/9, 11.1%) involved deep to the muscle (Figure 4).

FIGURE 4.

A 30‐year‐old man with a lobulated‐morphology cellular dermatofibroma (CDF) at his left upper arm with a size of about 4.6 × 3.3 × 1.9 cm³, who had undergone a CDF resection within 19 months. A new mass in the same location was found at 5 months before his admission to our hospital. (A–C) The tumor showed several nodules fusing together with mixed intensity in T₁WI (A) and T₂WI (B); complete capsules could be observed (arrows). An obvious enhancement pattern was dominant (C). The basal area of the mass had a poorly‐defined boundary with muscle (triangular arrows). (D) Hemosiderin deposition could be displayed (notable hypo‐intensity) in enhanced coronal fat saturation (FS)‐T₁WI (triangle). (E) Time‐signal intensity curve (TIC) of different regions were respectively types III and Ⅱ, which could help to guide biopsy area. (F) Diffuse, uniform spindle cells were arranged in a storiform pattern in cellular regions (hematoxylin and eosin [H&E], x100)

FIGURE 5.

A patient with a cellular dermatofibroma (CDF) in the upper arm. Horn‐like protrusions within the subcutaneous fat and dermis could be observed (arrow), which was different from dermatofibrosarcoma protuberans (DFSP) but indicating the necessity of certain degree of wide local excision (WLE)

None of patients with keloids were found recurrence. All keloids (12/12, 100%) presented iso‐intensity in T₂WI, while diverse intensity in T₁WI (four cases [33.3%] with hyperintensity and eight cases [66.7%] with iso‐or‐hypo‐intensity), most of which (11/12, 91.7%) presented as flat spindles protruding from the epidermis, stretching along the short axis of the body if located in trunks. All 12 keloids (12/12, 100%) had clear margins and homogeneous enhancement pattern (Figure 6), of which 10 cases (10/12, 83.3%) had a mild‐to‐moderate enhancement, and two cases (2/12, 16.7%) had a significant enhancement. HR‐MRI showed 10 keloids (10/12, 83.3%) were confined within the dermis layer, and two keloids (2/12, 16.7%) involved the subcutaneous fat because of original deep skin trauma. Keloids had the most superficial depth (4.8 ± 1.7 mm, p = 0.04, Table 1).

FIGURE 6.

A 24‐year‐old female who was pathologically diagnosed with keloids at the lower abdominal wall, with a size of about 3.3 × 1.7 × 1 cm³. (A) Appearance of the neoplasms. (B–C) The keloids were shown as iso‐intensity in T₁WI (B) and T₂WI (C). (D) Enhanced T₁WI showed an enhanced protuberant nodule, which has a clear distinction from subcutaneous fat, stretching along peripheral dermis. The peripheral fatty tissue was compressed. (E) Type‐II time‐signal intensity curve (TIC) of the oval regions of interest (ROIs) in C. (F) Collogen bundles could be observed (hematoxylin and eosin [H&E], x40)

The NF presented as an irregular mass that grew along the deep fascia, with iso‐intensity in T₁WI and hyperintensity in T₂WI, further with multiple angular protrusions (mainly stretched along the fascial space). A relatively well‐defined margin and a homogeneous, significant enhancement pattern could be observed. Exudation of the peripheral fascia and multiple abnormal lymph nodes could be clearly visualized (Figure 7).

FIGURE 7.

A 44‐year‐old woman with a rapidly growing subcutaneous nodule at her left‐sided neck for 6 months, measuring approximately 2.8 × 1.4 × 1.6 cm³, was pathologically diagnosed with nodular fasciitis. (A–C) High‐resolution magnetic resonance imaging (HR‐MRI) showed an irregular mass with iso‐intensity in T₁WI (A) and hyper‐intensity T₂WI (B), growing along the deep fascia. The black arrow in B presented the thickened deep fascia. An obvious homogeneous enhancement pattern could be observed in C. (D and F) Fascia exudation (white arrow) and abnormal lymph nodes (triangles) could be observed around the mass. And protrusions mainly stretched along fascial space (triangle arrow). (E) Regions of interest (ROIs) in C had type‐Ⅱ time‐signal intensity curve (TIC)

3.3. Analysis of DCE‐associated quantitative parameters and TICs

The differences of all four parameters among DFSP, DF, and keloids were statistically significant (Table 2). DFSPs had the highest values of K ^trans, K_ep , V_{e ,} and iAUC (There were no significant difference of V_e values between DFSP and keloids [p = 0.059]) (Figure 8). Moreover, DFs had greater values of K_ep and iAUC than those of keloids (p = 0.026, 0.031).

TABLE 2.

Comparison of dynamic contrast‐enhanced (DCE)‐associated quantitative parameters

DCE‐associated parameters	Dermatofibrosarcoma protuberans (n = 7)	Dermatofibroma (n = 9)	Keloid (n = 12)	F‐values	p‐Values	Nodular fasciitis (n = 1)
K trans (min−1)	1.01 ± 0.56ab	0.20 ± 0.1a	0.10 ± 0.05b	11.49	0.002	0.42 ± 0.09
Kep (min−1)	1.58 ± 0.83ab	0.54 ± 0.31ac	0.20 ± 0.09bc	13.66	0.002	0.64 ± 0.07
Ve (min−1)	0.70 ± 0.14a	0.41 ± 0.09a	0.53 ± 0.18	7.52	0.003	0.64 ± 0.06
iAUC	44.87 ± 17.97ab	18.69 ± 8.87ac	8.96 ± 5.2bc	15.408	0.001	27.79 ± 7.37

Note: Data are all presented as mean ± SD. The same superscripts a, b, or c within the same parameter indicate that there was a significant difference between the both groups (p < 0.05).

FIGURE 8.

(A–D) Patient A, a 31‐year‐old male, with a dermatofibrosarcoma protuberans (DFSP) in the right supraclavicular fossa for 2 months. The lesion presented as hyper‐intensity in T₂WI, with irregular morphology. The base of the lesion was close to the underlying muscle, and the fat space disappeared (triangular arrow), which indicating deep fascia involvement. Time‐signal intensity curve (TIC) of regions of interest (ROIs) in B was type‐III. CD34 staining was strongly positive (D). (E–H) Patient B, a 53‐year‐old female, with a nodule in the right upper arm for more than 10 years, which growing rapidly accompanying with pain for 1 year. The pathology indicated a result of cellular dermatofibrosarcoma (CDF). The lesion showed as iso‐intensity in T₂WI, which was limited within the superficial subcutaneous fat. TIC of ROI in F was type III. CD34 staining was spotty positive. The dynamic contrast‐enhanced (DCE)‐associated quantitative parameters of patient A were higher than those of patient B

Six cases (6/7, 85.7%) and one case (1/7, 14.3%) of DFSP presented as type‐III and type‐I (Figure 2) TICs, respectively. All four cases of CDFs (4/4, 100%) showed type‐III TICs (Figures 4 and 9). In contrast, conventional and keloidal type of DFs, all keloid and the NF showed type‐I or type‐II TICs (Figures 6, 7, and 9). Significant differences were found in TIC types between A and B groups (p < 0.001) (Table 3).

FIGURE 9.

Two patients who had undergone surgery but found new nodules in original incisions. (A–B) A 36‐year‐old man who underwent a cellular dermatofibrosarcoma (CDF) resection 2 years before his admission to our hospital. He was diagnosed with a keloid after surgery. T₁WI showed its flat morphology confined within dermis (A). Type‐I time‐signal intensity curve (TIC) could be seen (B). (C–D) A 34‐year‐old woman who underwent resection of a CDF 1 month prior to admission to the hospital. A new subcutaneous nodule was found. T₁WI showed an oval nodule with iso‐intensity (C). TIC of the regions of interest (ROIs) in C was type‐III (D)

TABLE 3.

Comparison of TIC types

	Group A (n = 11)		Group B (n = 18)
Type of TIC	Dermatofibrosarcoma protuberans (n = 7)	Cellular dermatofibroma (n = 4)	Other DF variants (n = 5)	Keloid (n = 12)	Nodular fasciitis (n = 1)
I	1 (14.3%)	0	2 (40%)	10 (83.3%)	0
II	0	0	3 (60%)	2 (16.7%)	1 (100%)
III	6 (85.7%)	4 (100%)	0	0	0
p‐Value	<0.001

Abbreviations: DF, dermatofibrosarcoma; TIC, time‐intensity curve.

3.4. Histopathological findings

All lesions mentioned above showed with a diffuse spindle cell morphology under light microscopes. Monomorphic spindle cells with increased mitotic activity, arranged as storiform or fascicular patterns and few collagen bundles could be observed in DFSP lesions. The tumor cells infiltrated into subcutaneous adipose tissue in a honeycomb pattern (Figures 1H and 2G). Besides, CD34‐diffusely‐positive (6/7, 85.7%) (Figures 2H and 8D) and CD68‐negative (5/7, 71.4%) were identified in DFSP cases (one case had no IHC results due to the absence of IHC data). Furthermore, CDFs appeared as masses or nodules with extension between subcutaneous adipocytes, but without infiltration of the fat lobules, which were composed of highly uniform spindle cells arranged in a spoke wheel pattern. Touton giant cells, hemosiderin (Figure 4D), and collagen bundles could be observed. CD68 staining was positive (4/4, 100%), and CD34 staining was spotty positive (3/4, 75%) or negative (1/4, 25%) (Table 4). The conventional DFs were composed of bland spindle cells that arranged in a vortex pattern inlaid in collagen bundles. Regarding keloidal‐type DF, densely arranged hyaline collagens dyed in red homogeneously were partly observed. As for keloids, massive fibroblasts and thick bundles of collagen fibers could be seen (Figure 6F). The NF was composed of uniform spindle cells that arranged in interwoven pattern. Extravasated erythrocytes and infiltrated inflammatory cells could be visualized. Lymph nodes around the lesions presented with signs of chronic inflammation. Vimentin‐positive and SMA‐positive, as well as CD34‐negative, were detected. The involvement layers judged by histopathology are listed in Table 4.

TABLE 4.

Population, immunohistochemistry, pathology, and high‐resolution magnetic resonance imaging (HR‐MRI) features of DFSPs and CDFs

Number	Pathology	Gender	Age	CD34	CD68	KI‐67(%)	Pathology involvement layer	Peripheral infiltration in HR‐MRI
1	DFSP	31	Male	+	–	10+	Deep fascia	No
2	DFSP	49	Female	+	–	10+	Subcutaneous fat	Yes
3 †	DFSP	66	Male	/	/	/	Deep fascia	No
4	DFSP	33	Female	+	–	10+	Subcutaneous fat	Yes
5	DFSP	24	Male	+	+ ‡	15+	Deep fascia	Yes
6	DFSP	66	Male	+	–	30+	Subcutaneous fat	Yes
7	DFSP	25	Male	+	–	3	Subcutaneous fat	Yes
8	CDF	30	Male	–	+ ‡	15+	Muscle	No
9	CDF	53	Female	+ ‡	+ §	5+	Subcutaneous fat	No
10	CDF	34	Female	+ ‡	+	20+	Subcutaneous fat	No
11	CDF	39	Female	+ ‡	+ §	5+	Subcutaneous fat	No

Abbreviations: CDF, cellular dermatofibroma; DFSP, dermatofibrosarcoma protuberans.

^† This patient had no immunohistochemistry (IHC) results due to absence of IHC data.

^‡ The IHC staining was positive spotty.

^§ The IHC staining was positive scatteredly.

4. DISCUSSION

To our knowledge, preoperative accurate diagnosis of DFSP is highly vital, which determines the way to excise and prognosis. A notable advantage of HR‐MRI in the diagnosis of DFSP is that fat saturation (FS)‐sequence images (especially FS‐enhanced‐T₁WI) could clearly show the infiltration of peripheral adipose tissue (Figure 3) due to its HR feature, and elimination of fat tissue that has significant hyperintensity. This is crucial for the diagnosis of DFSP, ⁵ as this can suggest the invasive growth, as well as the spread along the septa of subcutaneous adipose tissue of tumor cells. And this infiltration feature in HR‐MRI may correspond to the honeycomb‐pattern infiltration of spindle cells at the histopathology level, which is the reason for the ease of recurrence of DFSP after SE and the histopathological basis for the implementation of WLE and MMS. Also, the fat infiltration is the key to differential diagnosis of DFSPs and CDFs, due to the tumor cells of CDFs can also extend into the subcutaneous fat, but do not infiltrate the fat septum and lobules. ¹⁰ In this present research, there was one case of DFSP with tiny spinous protrusion lengthened up to 1.29 cm (Figure 3A), which highlighted the need for WLE. While some cases only displayed infiltration in the basal of lesions (Figure 3F), in which the benefits from WLE remains to be proved.

DFSPs are most easy to confuse with CDFs, especially those involved subcutis. Three cases (3/4, 75%) of CDFs had vague margins in this research, but no fat infiltration symptoms like those of DFSPs were found. Previous researches ⁵ , ¹⁰ indicated that WLE was also efficient for CDFs, but with narrower margin than DFSP, which was identical to manifestation of HR‐MRI (Figure 5). Complete capsules also imply noninfiltrative growth of the CDF (Figure 4). Keloids also need to be differentiated from DFSPs, especially when the lesion is notably protruded from the skin surface. HR‐MRI could show keloids’ well‐defined margins, as well as superficial location (Figure 6), which was obviously different from DFSPs.

DCE‐MRI possesses a unique capability to quantitatively measure the microcirculation in living tissues. ²³ As mentioned earlier, DFSP cases had the highest values of K ^trans and K_ep , which reflected increased neovascularization, immaturation, and damages of vascular endothelial cells within DFSP lesions ²⁷ , ²⁸ (Figure 1G). The V_e values are positively correlated with the degree of tissue necrosis. ²⁹ The V_e value of DFSPs was higher than that of DFs. The reason might be the more microscopic necrosis in DFSPs than DFs, which are often associated with dense tumor cells in malignant lesions. There was no significant difference of V_e values between DFSPs and keloids, which might result from the insufficient sample size. DFSPs had the largest iAUC value, which indicated that DFSPs had the highest degree of enhancement, densest tumor cells, and richest blood supply. Generally, DFSPs, DFs, and keloids showed a decreasing value of iAUC (i.e., increasingly lighter enhancement degree ³⁰ ), which was consistent with a decreasing proportion of cellular components.

Although DFSP and CDF both have abundant cells and active proliferation, values of K ^trans, K_ep , and V_e of DFSP were still higher than those of CDF cases (mean values: 1.01, 1.58, 0.70 vs. 0.30, 0.79, and 0.41, respectively). When imaging features are not adequate to reach a robust conclusion, quantitative parameters could be referred (Figure 8). Besides, CDFs are also more prone to be confused with other stable variants of DFs. CDF cases had greater values of K ^trans and K_ep than those of other variants (mean values: 0.30, 0.79 vs. 0.12, 0.35).

In the present research, as shown in Table 3, there was an extremely significant difference in TIC type between both groups. Previous researches showed that lesions with type‐III TICs were indicative of a high‐risk of malignance. ²⁴ , ²⁵ DFSPs and CDFs both had rich cellular components and active mitosis, indicating abundant oncological neovascularization. This might be the reason for the rapid clearance of the contrast agent in the lesions and the appearance of type‐III TICs. There was one DFSP case, which showed type‐ⅠTIC. Significantly more collogen bundles could be observed within the lesion (Figure 2F), which might be responsible for type‐ⅠTIC. Conventional type and keloidal type DFs, as well as all keloids and the NF all presented type‐Ⅰor type‐Ⅱ TICs, which made it possible to take advantage of type‐III TICs to identify DFSPs and CDFs from other lesions. To date, many studies had been conducted on DFSPs, while there was an obvious lack of research on CDFs. The use of TICs might help direct clinicians’ attention to the treatment methods and outpatient follow‐up for patients with CDFs. What's more, the characteristics of different regions of the lesion could be presented according to TIC type (Figure 4E), which might be used to guide biopsy of specific areas in heterogeneous tumors.

Another advantage of HR‐MRI is related to its integrally imaging capability, especially for tumors invading deep fascia and muscles, where exact depth values can be easily measured ³¹ with little distortion of peripheral anatomy. This capability is significant when the excision depth needs to be determined. Meanwhile, the details of lesions would not be missed. The infiltration of DFSPs to the surrounding fat (Table 4) could display with high resolution, which provided strong support for the implementation of WLE. In contrast, conventional cutaneous imaging methods, such as high‐frequency ultrasound, dermoscopy, and confocal microscopy ³² can hardly display images with the above‐mentioned features simultaneously. Conventionally, the final diagnosis of DFSPs depends on needle biopsy, which is invasive, time‐consuming, and occasionally inaccurate. Further, needle biopsy cannot reliably indicate involvement layers. This also suggests the possibility of HR‐MRI as a complementary or even alternative method.

Our study had several limitations. Firstly, due to the short‐term follow‐up, no recurrent case was detected. Secondly, no precise cut‐off values of quantitative parameters for differential diagnosis were provided owing to the small sample size. Finally, manual placement of ROI may cause bias in quantitative parameter measurements.

5. CONCLUSIONS

In conclusion, DCE‐HR‐MRI could show the growth characteristics of DFSPs, which was of high value for the diagnosis and differential diagnosis of DFSPs and was helpful for the determination of treatment options, thereby to improve the prognosis of patients.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Yu Q, Zhu Y, Huang R, Li Y, Song L, Zhang X, et al. Diagnosis and differential diagnosis of dermatofibrosarcoma protuberans: Utility of high‐resolution dynamic contrast‐enhanced (DCE) MRI. Skin Res Technol. 2022;28:651–663. 10.1111/srt.13164

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.