Ethical approval
This study was conducted in accordance with the principles embodied in the Declaration of Helsinki. This study was also approved by the Ethics Committee of Kyushu University (No. 23177-00). Written informed consent was obtained from each patient before his/her enrolment to the research.
Materials
Samples from 68 cases between 2004 and 2023 that were previously diagnosed as DFSP were retrieved from the soft-tissue tumors registered in the files of the Department of Anatomic Pathology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan. Among them, 21 cases with fibrosarcomatous (FS) lesions were enrolled in this study, while we also collected 21 cases of age/sex/location-matched DFSP as a control group.
Immunohistochemistry
From formalin-fixed, paraffin-embedded blocks, a single representative block was chosen for each case for immunohistochemistry (IHC) and was sectioned at 3 mm thickness. H&E staining and IHC staining for CD34 (QBEnd10; Dako, Glostrup, Denmark) and PRAME (ER20330; Abcam, Cambridge, UK) were performed. The primary antibodies, their dilutions, and the antigen retrieval are summarized in Supplementary Table 1. As positive and negative controls for PRAME, malignant melanoma cases and common nevus cases were selected, respectively. The immune complexes were detected with the DAKO EnVision Detection System (Dako).
We then performed double immunostaining of PRAME and CD34. In addition to the immunodetection using the horseradish peroxidase method mentioned above, N-Histofine Simple Stain AP (Nichirei Biosciences Inc., Tokyo, Japan) and PermaBlue Plus (Diagnostic BioSystems Inc., Pleasanton, CA) were used as the secondary antibody and the chromogen, respectively.
H-scoring and imaging analysis
IHC for CD34 was semi-quantitatively evaluated, using the H-score. In each high-power field (HPF) (´400), the positivity of each cell was classified into the following four classes: negative: 0, weakly positive: 1, moderately positive: 2, and strongly positive: 3. We obtained the H-score (0–300) by multiplying the proportion (%) of stained cells by the staining intensity (0–3). We selected three random fields from the tumor lesion and calculated the average of their H-scores as the final H-score. We used QuPath ver. 5.1 [15], an open-source platform, for bioimaging analysis. We created projects by importing JPEG images of the HPFs and performed positive cell detection to recognize the proportion and intensity of the IHC. The nuclear parameters were set as follows: nuclear size of 150 to 2000 pixels2, cell expansion of 20 pixels, and background radius of 40 pixels. We did not split the image by shape. As for the intensity threshold parameters, we evaluated the maximum staining in the cytoplasm and divided it into three threshold levels: weak (+1, highlighted in yellow), moderate (+2, orange), and strong (+3, red). Negatively stained cells were highlighted in blue. QuPath software automatically provided us with an H-score.
As for the evaluation of IHC for PRAME, because the staining intensity varied markedly within a tumor lesion, we could not obtain an H-score by simply averaging HPFs from random areas. Therefore, we classified the whole slides into four areas (negative, weakly positive, moderately positive, strongly positive) via observations at low magnification (´100), and calculated the percentage of each area among the whole slide (area; %). Then, we obtained the H-score/HPF for each area, in the same way as for CD34, using QuPath. As for the intensity threshold parameters, we evaluated the maximum staining in the nucleus. We calculated the final modified H-score (mH-score) by summing up the scores of the percentage of each area multiplied by the H-score/HPF of the area [negative area (%) × H-score/HPF (negative) + weakly positive area (%) × H-score/HPF (weakly positive) + moderately positive area (%) × H-score/HPF (moderately positive) + strongly positive area (%) × H-score/HPF (strongly positive)] (Supplementary table 2).
Statistical analysis
We statistically analyzed the obtained H-score and mH-score using the Wilcoxon singled-rank test in the statistical analytical software JMP (ver. 17.0.0) (Supplementary table 3). A two-sided p-value of <0.05 was considered to be statistically significant.