This retrospective study analyzed 13 intimal sarcoma (IS) cases and 35 angiosarcoma (AS) cases identified from pathology archives between January 2017 and April 2025. Formalin-fixed paraffin-embedded (FFPE) samples from surgical resections or biopsies were collected for next-generation sequencing (NGS), immunohistochemistry (IHC), and fluorescence in situ hybridization (FISH) when available. Clinical data were retrieved from hospital records, and patients were followed for survival outcomes. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Review Board of the China-Japan Friendship Hospital (2023-KY-045). Individual consent for this retrospective analysis was waived.

Given the rarity of IS, we supplemented our cohort with 18 IS patients from a publicly available dataset (Memorial Sloan Kettering Cancer Center, MSKCC) as an independent cohort. The genomic data (cancer panel) and clinical information for these 18 cases were obtained via cBioPortal (44).

Surgical and biopsy specimens underwent formalin fixation and paraffin embedding, followed by sectioning at 4.0 µm thickness. Sections were stained with hematoxylin and eosin (H&E) and immune-stained using the EnVision method, incorporating both negative and positive controls.

Primary antibodies were sourced as follows: CKpan (AE1/AE3), Vimentin (VMAB159), CD34 (EP88), Ki-67 (MIB-1), CD8 (SP16), and CD4 (EP204) from Beijing Zhong Shan - Golden Bridge Biological Technology Co., Ltd. (Beijing, China); SMA (1H4), CD3 (MX036), CD68 (Kp-1), MDM2 (IF2), and CDK4 (EP180) from Fuzhou Maixin Biotechnology Development Co., Ltd. (Fuzhou, China); PD-L1 (22C3) from Dako North America, Inc.; CD3 (L26) along with all secondary antibodies and reagents from Roche Diagnostics (Shanghai) Co., Ltd. Protein localization analyses were performed using established subcellular compartment criteria: nuclear staining for MDM2 and Ki-67; cytoplasmic staining for SMA, CKpan, CD3, CD8, CD4, CD68, and Vimentin; exclusive plasma membrane staining for PD-L1; combined cytoplasmic and plasma membrane staining for CD34; and cytoplasmic/nuclear dual localization for CDK4.

Immunostaining was scored as (+) when 1-10% of tumor cells were positive, (++) for 10% to 40% positive cells, and (+++) when positivity exceeded 40%. PD-L1 expression was assessed using the Tumor Proportion Score (TPS), defined as the percentage of viable tumor cells exhibiting partial or complete membrane staining at any intensity, with a positivity cutoff set at 1%.

The MDM2 (12q15) gene probe was employed to assess corresponding gene amplification in tumor cells using fluorescence in situ hybridization (FISH). The FISH probe kit was acquired from Guangzhou Amping Pharmaceutical Technology Co. Ltd. Procedures were performed in strict accordance with the manufacturer’s protocol. For interpretation, the average signals for MDM2 and CEP12 were enumerated across 50 tumor cells. MDM2 amplification was defined as an MDM2/CEP12 signal ratio ≥ 2.0.

Genetic analysis was performed as previously described (45). In brief, serial sections of formalin-fixed paraffin-embedded (FFPE) tumor tissues were subjected to genomic DNA extraction using the QIAamp DNA Mini Kit (Qiagen, Valencia, CA). Histologically confirmed adjacent noncancerous tissue served as the control. Sequencing libraries were prepared using Illumina TruSeq DNA Library Preparation Kits (Illumina, San Diego, CA). Libraries were hybridized with custom-designed biotinylated oligonucleotide probes targeting either 1021 genes or the whole-exome sequencing (WES) panel (Integrated DNA Technologies, Inc.; Supplementary Table 1 ). Final libraries were sequenced on the GenePlus-Seq-2000 platform (GenePlus-Suzhou Institute).

The sequencing data were analyzed using default parameters (45). Adapter sequences and low-quality reads were removed. Clean reads were aligned to the human reference genome (hg19) using the Burrows-Wheeler Aligner (BWA; version 0.7.15-r1140). Single nucleotide variants (SNVs) were called using MuTect (version 1.1.4) and NChot. Small insertions and deletions (Indels) were identified with GATK. Somatic copy-number alterations (SCNAs) were detected using CONTRA (version 2.0.8). Significant SCNAs were defined as the log2 ratio of adjusted depth between tumor DNA and matched germline control DNA. All final candidate variants underwent manual verification in the Integrative Genomics Viewer (IGV). Targeted capture sequencing required a minimum mean effective depth of coverage of 300× for whole-exome sequencing (WES) and 500× for the 1021-gene panel. Variants were filtered to exclude synonymous alterations, known germline variants listed in dbSNP, and variants with a population frequency exceeding 1% in the Exome Sequencing Project.

We performed RNA-Seq as previously described (46). Briefly, total RNA was extracted from tumor FFPE specimens using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. RNA integrity was assessed using the RNA Integrity Number (RIN) generated by the 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA libraries were constructed with the NEBNext^® Ultra™ RNA Library Prep Kit (New England Biolabs, Beverly, MA, USA). Libraries were sequenced on the Geneplus-2000 platform (Geneplus, Beijing, China).

The relative fractions of 22 infiltrating immune cell types within each tumor tissue were determined using CIBERSORT (http://cibersort.stanford.edu/). The algorithm was executed using the LM22 signature matrix with 1,000 permutations. The evaluated tumor-infiltrating immune cell populations included naïve B cells, memory B cells, plasma cells, naïve CD4⁺ T cells, resting memory CD4⁺ T cells, activated memory CD4⁺ T cells, follicular helper T cells (Tfh), regulatory T cells (Tregs), CD8⁺ T cells, gamma delta T cells (γδ T cells), M0 macrophages, M1 macrophages, M2 macrophages, resting natural killer (NK) cells, activated NK cells, resting dendritic cells (DC), activated DC, monocytes, resting mast cells, activated mast cells, neutrophils, and eosinophils. For each tumor sample, the sum of the fractions for all evaluated immune cell types equaled 1 (47, 48).

The cohort comprised 66 cases (34 females, 32 males) with a mean age of 50.5 years (range, 17–83 years). The IS subgroup exhibited a higher proportion of female patients (20/31) compared to the AS subgroup (15/35). Tumor primary sites differed significantly, with IS predominantly involving the pulmonary artery and heart, and AS primarily involving the skin ( Table 1 ).

As pathological information for the 18 IS patients from the MSKCC cohort was unavailable (44), we focused on the 11 IS patients with detailed pathological analysis. Histologically, all tumors exhibited moderately to severely atypical spindle tumor cells. Epithelioid or giant tumor cells were present in 7 cases. Among the 8 cases with necrosis, these areas consistently featured severely atypical tumor cells, including epithelioid and bizarre giant tumor cells. Tumor cell density varied regionally. Some areas showed relatively sparse tumor cells with stromal myxoid degeneration, while others displayed densely packed spindle cells forming fascicles reminiscent of leiomyosarcoma. One case exhibited tumor invasion accompanied by destruction of the vascular wall. Osteosarcoma differentiation was identified in some cases ( Figure 1A ).

The tumor cells of IS and AS were diffusely positive for Vimentin (11/11, 100%) but negative for CKpan. Some IS tumor cells expressed MDM2 (8/11, 72.7%), SMA (3/11, 27.3%), and CD34 (2/11,18.2%). The Ki-67 proliferation index ranged from 30% to 90%, indicating a relatively high proliferation rate ( Figures 1B, C ).

Archived FFPE samples were utilized for tumor genomic DNA and RNA extraction when available. For the 48 patients from our center, all provided sufficient DNA for subsequent NGS analysis, with 5 undergoing whole-exome sequencing (WES) and the remaining 43 analyzed using a 1021-gene panel. The 18 IS patients from the MSKCC cohort underwent sequencing with IMPACT341/410/468 panels. Additionally, qualified RNA was obtained from 17 patients (11 IS and 6 AS) for RNA sequencing.

Somatic mutations were identified in all 66 patients, with 522 single nucleotide variants (SNVs) detected in the 31 IS patients and 518 SNVs in the 35 AS patients. Copy number variants (CNVs) totaled 268 in IS patients and 65 in AS patients. Collectively, the most frequently mutated genes were MDM2 (36%, 24/66), CDK4 (27%, 18/66), TP53 (26%, 17/66), KIT (23%, 15/66) and PDGFRA (20%, 13/66) ( Figure 2A ).

TP53 mutations were significantly more prevalent in AS patients compared to IS patients (15/35 vs 2/31, p < 0.001, Figure 2B ). Conversely, amplifications of KDR, KIT and PDGFRA on chromosome 4, as well as CDK4 and MDM2 on chromosome 12, were enriched in IS patients (p < 0.001; Figure 2C ).

Moreover, CDK4 copy number gains or CDKN2A/CDKN2B losses were observed in 25 of 31 IS patients but only in 2 of 35 AS patients (p < 0.001; Figure 2C ). This cell cycle dysregulation may underlie the elevated Ki-67 index 30%- 90% in IS patients ( Figure 1B ).

Overexpression and amplification of MDM2 constitute an important characteristic in IS. Further analysis was performed on IS patients lacking MDM2 CNV within our cohort. Among the four patients with available MDM2 IHC staining results, two were MDM2 IHC ++ (P003, P007) and two were MDM2 IHC + (P002, P010). Subsequent MDM2 FISH analysis on samples P007 and P010 also showed no MDM2 copy number gain in either specimen. ( Figure 2D , Supplementary Table S1 ). The observed discrepancies among IHC, NGS-CNV, and FISH results are likely attributable to the low tumor cell fraction in these samples.

Given the striking differences in genomic profiling between the IS and AS cohorts, we further investigated characteristics of the tumor microenvironment in these groups. Using the CIBERSORT algorithm to calculate the abundance of 22 immune cell types within each sample, we found no significant difference in immune infiltration between the IS and AS cohorts ( Figure 3A ).

We next examined PD-L1 expression via IHC in IS patients. Among eight patients with sufficient FFPE samples, PD-L1 expression was detected in six, with three patients (P002, P004, P007) exhibiting strong positivity (70%, 80%, and 70%, respectively; Figure 3B , Supplementary Figure S1 ). Although the small sample size (n=8) limits broad conclusions, this preliminary finding suggests that PD-L1 expression is a clinically relevant feature in a considerable proportion of IS patients and may indicate a potential for response to ICB. Given the observed abundant immune cell infiltration- including lymphocytes, plasma cells, and histiocytes, with some lymphocytes forming stromal clusters- we characterized the immune microenvironment using the following markers: CD3+ for total T cells, CD4+ for helper T cells, CD8+ for cytotoxic T cells, CD68+ for macrophages, and CD163+ for monocytes/macrophages. Analysis revealed median immune cell infiltration levels of 13% CD3+ (range, 1–80%), 10% CD8+ (range, 2–75%), 30% CD4+ (range, <1–70%), 70% CD68+ (range, 20–90%), and 80% CD163+ (range, 30–90%) ( Figure 3B , Supplementary Figure S2 ). Notably, patients who died within 12 months post-surgery (n=4, P003, P005, P008, P010) exhibited significantly lower CD3+ infiltration compared to those surviving beyond 12 months (n=3; P001, P004, P007; median 7.5% vs. 70%, P=0.029; Figure 3C ).

Among the 13 IS patients treated at our center, eight underwent pulmonary endarterectomy and one received wedge resection of the left lower lobe. Three patients died postoperatively on day 5, month 4, and month 11, respectively. Of the five surviving surgical patients, two maintained disease-free survival at 11–14 months. Two experienced relapse, while one patient relapsed at 12 months but remained alive for over 56 months with maintenance therapy comprising sintilimab (anti-PD-1 immunotherapy) and anlotinib (targeted therapy). Two patients did not undergo pulmonary artery tumor resection: one due to inoperability (adrenal metastasis) and another who refused surgery and succumbed to disease progression at 5 months ( Table 2 ).

Three patients received immune checkpoint inhibitor (ICI) therapy during their disease course. Patient P001, an adult patient with pulmonary artery intimal sarcoma involving the pulmonary trunk (PT), left pulmonary artery (LPA), and right pulmonary artery (RPA), received adjuvant sintilimab following pulmonary endarterectomy (PEA) and has remained progression-free for over 14 months to date. The tumor exhibited 15% PD-L1 expression and strong MDM2 positivity, with an MDM2 copy number gain (CNG) of 9.15. Patient P007, an adult patient also diagnosed with pulmonary artery intimal sarcoma involving the PT, LPA, and RPA, experienced tumor recurrence 12 months post-PEA. This patient subsequently underwent 4 cycles of combined therapy including chemotherapy, the antiangiogenic agent anlotinib, and sintilimab, followed by sintilimab maintenance therapy, achieving sustained tumor control for over 56 months since initial diagnosis ( Figure 4 ). Patient P005, an adult patient with pulmonary artery intimal sarcoma confined to the LPA, underwent wedge resection of the left lower lobe. Despite negative PD-L1 expression, adjuvant therapy comprising combined chemotherapy, antiangiogenic therapy, and an ICI was administered; however, the patient died 11 months post-surgery. The tumor was MDM2-positive, with an MDM2 CNG of 5.37.