The patient was a 56-year-old Chinese man accidentally detected a kidney mass by abdominal ultrasound in outer hospital due to repeated hiccups in August, 2019. Then the patient was referred to our hospital for further diagnosis and treatment in September, 2019. No family tumor history was mentioned and no clinical symptoms were presented, such as flank pain, fever and gross hematuria, nevertheless his weight had lost about 5 kilograms since the onset. Physical examination found that multiple subcutaneous soft nodules located in head, face, neck, waist, abdomen, chest, back and limbs with the diameter from 0.8-1.5cm, without redness, swelling, ulceration and purulence. The enhanced computed tomography (CT) of abdomen revealed an uneven-density mass in the upper pole of the right kidney and multiple uneven-density nodules in abdominal cavity, abdominal wall, retroperitoneum, pelvic cavity, bilateral groin, bilateral psoas muscles, iliopsoas and gluteal muscles, front abdominal wall, back and hip subcutaneous and around the anus (Fig. 1A1, A4 and B4). The kidney mass was measured with the size of 46mm×42mm×39mm and shown low-density and no-enhanced necrotic area in portal phase (Fig. 1A2). Another low-density nodule was found in the lower pole of the right kidney with the size of 29.74 × 22.70 × 21.77 mm (Fig. 1A3). Chest enhanced CT showed no metastasis in the lungs, but there were multiple enlarged stuck nodules in the mediastinum, bilateral armpits and anterior chest wall, which was considered as metastases (Fig. 1B1-B3). A peritoneal metastatic node was removed for frozen-section (yellow triangle) (Fig.1C). Then, a right radical 3D laparoscopic nephrectomy plus peritoneal nodule biopsy was performed. The patient returned to the hospital for review 1 month after surgery and was admitted oral Pazopanib treatment (200mg per tablet, qd) on Oct 13th, 2019. During the treatment of Pazopanib, the patient was getting better physically. Unfortunately, the patient died 1 month later due to excessive medication arbitrarily. The entire course of his clinical treatment is illustrated in Supplementary Fig. 1.
Pathological findings
Macroscopic findings
Grossly, the tumor protruding from the renal cortex was confined within the renal parenchyma and measured to be 4.5 × 4.0 × 4.0 cm. The cross-section showed a yellowish and spherical neoplasm in the upper pole of the right kidney. The border with the kidney is sharp. Focal hemorrhage and necrosis were identified. In addition, a few dense gray-white nodes were detected in the renal capsule, the perinephric adipose tissue and peritoneum, and their diameter was about 0.3-1.0 cm (Supplementary Fig. 1).
Histopathological and immunohistochemical (IHC) findings
Microscopic evaluation of the renal tumor showed the tumor was multi-nodular and separated by fibrous tissue. The tumor histology was very complicated with three different histological morphologies, including conventional ccRCC like region, eosinophilic papillary structure and sarcomatoid (spindle) differentiation (Fig. 2A). In ccRCC like region, the pattern of growth was predominantly solid, with formation of nest and acinar patterns separated by the stroma that was characteristically endowed with a prominent network of small, thin-walled blood vessels (Fig. 2Aa1). The area of eosinophilic papillary structure was composed of papillae formed by delicate fibrovascular cores that contain foamy macrophages. Most of these tumor cells showed abundant eosinophilic cytoplasm with occasional small nuclei (Fig. 2Ab1). The sarcomatoid area was lack of distinct borders with the area of clear cells which transition zone between them could be seen. In this region, some little abscesses with lots of neutrophils and tumor necrosis were observed (Fig. 2Ac1). In addition, histopathological examination of the peritoneum node revealed epithelioid sheets of cells exhibiting great pleomorphism and frequent mitoses (Fig. 2B). These tumor cells with eccentric nuclei and abundant eosinophilic cytoplasm resembled rhabdomyoblasts, which was designated as rhabdoid differentiation (Fig. 2B1).
To identify the specific subtype of this tumor, we completed serious of IHC staining in the primary kidney tumors and peritoneal metastatic node. These immunostainings showed that CK8/18 and EMA were diffuse or patchy strong positivity in the classical ccRCC like area and eosinophilic papillary area (Fig. 2Aa2-3, 2Ab2-3), while their expressions were reduced in the sarcomatoid area and peritoneal metastatic node (Fig. 2Ac2-3, 2B2-3). To the opposite, vimentin was expressed more intensely and diffusely in the sarcomatoid area and peritoneal metastatic node (Fig. 2Aa4-Ac4, 2B4). Carbonic anhydrase Ⅸ (CA-Ⅸ) was present strongly and diffusely in the intact membranous distribution in the clear cells and those tumor cells in the eosinophilic papillary area, which was negative in sarcomatoid cells (Fig. 2Aa5-Ac5, 2B5). Tumor cells in these four lesions were negative for CD10, CK7, CD117, PAX8, MelanA, HMB45, SMA, Myogenin, Syn, CgA and ALK and retained SMARCB1 (INI-1) expression. The immunostaining summary information could be seen in Supplementary Table 1.
Tumor immune microenvironment (TIME) analysis of different regions
Due to numerous inflammatory cells accumulating in the sarcomatoid area and peritoneal metastatic node, we further detected the status of the immune tumor microenvironment via the IHC staining of LCA, CD8, CD4 and MPO and evaluated inflammatory cells in the ccRCC like region, sarcomatoid area and peritoneal metastatic tumor, respectively. We found that lymphocytes were mostly distributed in the border of different morphological tumor node, but not in the center of lesion, while the distribution of neutrophils had no zonal (Fig. 3A). The number of TILs infiltrating on the border of sarcomatoid area was significantly more than the ccRCC like region (p=0.0144, Fig. 3Aa2-b2 and 3C). Amount of CD8+ T cells in the sarcomatous area and peritoneal metastasis tumor were significantly more than ccRCC (psarcomatoid/ccRCC<0.0001, pperitonael / ccRCC<0.0001, Fig.3Aa-b4, 3B4 and 3E). The number of CD4+ T cells in the peritonaeul tumor was slightly higher than ccRCC area and sarcomatous area (pperitonael/ccRCC =0.0161, pperitonael/sarcomatoid =0.0161, Fig.3Aa3-b3, 3B3 and 3D). It implies that CD8+ T cells preferred infiltrating into the high grade RCC compared with CD4+ T cells. Moreover, neutrophils were more enriched in the sarcomatous areas and peritoneal tumor than ccRCC like region (psarcomatoid/ccRCC<0.0001, psarcomatoid/peritonae<0.0001, Fig. 3Aa5-b5, 3B5 and 3F).
Gene mutation analysis with target capture sequencing
To confirm the accurate subtype of tumor and further reveal the underlying molecular mechanism related heterogeneous morphology and extremely aggressive biological behavior, genomic profiling of the four distinct morphology tissue slides from resected tumor specimens was achieved by using target capture sequencing with a designed panel of 620 key cancer-related genes (GloriousMed Clinical Laboratory Co., Shanghai, China). The distribution of somatic mutations (SNV, InDel) among four different morphological tissues from this patient are summarized in a Venn diagram (Fig. 4A).
The most common VHL mutations in ccRCC were detected in all of them, including primary and peritoneal metastatic foci (Fig. 4B). Therefore, we identified this patient as a high-grade clear cell renal cell carcinoma (ccRCC) with with sarcomatoid differentiation and systemic multiple metastases (pT3aN1M1). Further analysis indicated tumor of eosinophilic papillary area carries the fewest number and types of mutated genes among all tumor tissues, only VHL and BAP1mutation, suggesting that these two genes might be early events in ccRCC tumorigenesis. From the perspective of genomic changes, eosinophilic morphology tumor seems to occur in the earliest stage of the disease due to the minimum number of driver gene alterations and mutation types. Notably, histone modification gene SETD2 mutations were detected in all of tumor regions except for eosinophilic papillary area, suggesting SETD2 could play an important role in ccRCC progression.
Remarkably, the genetic alterations in sarcomatoid regions are similar to those in clear cell carcinoma areas, both sharing some identical gene mutations such as VHL(p. R64fs), SETD2(p.N1628T), SMAD4(p.Q534fs*3), and PTPRT(p.A696V). BAP1 variant P190R and ATM variant E1199 deletion were observed only in conventional clear cell carcinomas, whereas there were exclusive TP53 variant A347G and PDGFRA variant S890 frameshift deletion in sarcomatoid-like tumor, hinting that TP53 and PDGFRA may play a significant role in the sarcomatoid transformation from ccRCC morphology. BRCA2 and ATR involved in DNA damage response pathway, NF-kappa B signaling gene CYLD and Hippo signaling gene YAP1, protein digestion and absorption gene COL5A3 were specifically detected in peritoneal metastatic node but not in other primary areas (Fig. 4B).
Besides, the tumor mutation burden (TMB) value of the four tumor areas present some certain regularities. The higher the grade, the higher the TMB value, as expected, the worse the prognosis, especially with sarcomatoid lesions. The TMB of peritoneal metastatic tumor was higher than the primary tumor (Fig. 4C). The TMB evidence also supports that tumor of eosinophilic papillary area and ccRCC area occurred at the early stage, then the tumor developed into the sarcomatoid morphology, and lastly metastasized to multiple organs including the peritoneum.
The last main thing is that this patient harbored several somatic mutations of certain genes including histone methyltransferase SETD2 (SET domain containing 2), ATR serine/threonine kinase (ATR), ATM serine/threonine kinase (ATM) and TP53 associated with DNA damage response signaling and Hippo signaling (YAP1), PI3K-Akt signaling (PDGFRA) and T cell receptor signaling (COL5A3) (Fig. 4B), some of which contribute to the development of the RCC in light of previous data [9, 17, 18].