Renal neuroendocrine tumors: clinical and molecular pathology with an emphasis on frequent association with ectopic Cushing syndrome

We identified surgically resected tumors diagnosed as primary renal neuroendocrine tumor or carcinoid over a period of 11 years (from 2011 to 2022) from the in-house surgical pathology files at the departments of pathology of the University Hospitals of Technical University Munich, University Hospital Düsseldorf, and University of Erlangen and also from the consultation files of two of the authors (AA and GK). NETs that had metastasized from other organs to the kidney were excluded in all cases by radiological and clinical findings. In all cases, formalin-fixed paraffin-embedded (FFPE) tissue blocks were available. The selected cases were reviewed and classified according to the WHO standards [48]. Nuclear Ki67 labeling was counted in more than 500 tumor cells in the area with the highest density (hot spot), and its percentage was reported as Ki67 index. Data on age, sex, tumor size and site, other renal diseases, hormonal symptoms, and presence or absence of metastasis were extracted from the available documents (see Table 1). This study was approved by the ethics committee of the Technical University Munich, Germany (approval number 2022–396-DFG-SR).

Immunohistochemical staining was performed on 2-µm sections using an automated system (Benchmark XT; Ventana/Roche, AZ, USA). Details regarding the immunohistochemical stainings are given in Supplementary Table 1. The expression was regarded as diffuse when all tumor cells were strongly and evenly stained or patchy when the staining of the tumor cells alternates between weak and strongly and the weakly stained cells dominated. A single cell positivity (< 5%) was regarded as negative. Membranous expression of the somatostatin receptor 2 (SST2) was evaluated based on the previously described method and a score 2 + and score 3 + were regarded as positive [42].

DNA and RNA, respectively, were each extracted from 5 slides with 8-µm sections of FFPE materials that contain tumor tissue more than 50% of the section area. After proteinase k digestion, nucleic acids were extracted using a semi-automated extraction system (Maxwell RSC 48, Promega, Madison, USA). Nucleic acid quantity was fluorometrically measured using the DNA high-sensitivity kit or the RNA high-sensitivity kit and the QuBit 4.0 instrument (Thermo Fisher Scientific, Waltham, USA). The amount of amplifiable DNA (sequencing grade quality) was determined using a commercially available qPCR assay (TaqMan RNAse P detection assay) while RNA quality was determined by a custom-designed qPCR approach for amplification of the RNA of a housekeeping gene (HPRT) on a StepOnePlus instrument (Thermo Fisher Scientific). The DNA amount as input for library preparation was adjusted according to the amplifiability of the DNA and the grade of degradation. Hybrid capture and library preparation were conducted using the TruSight Oncology 500 assay (Illumina, San Diego, CA, USA), following the manufacturer’s protocol. This assay allows targeted-capture sequencing of 523 cancer-related genes on the DNA level and translocation detection of 50 driver fusion genes on the RNA level. Up to eight DNA/RNA pairs were pooled together and sequenced on a NextSeq 550DX (Illumina) system using a NextSeq 500/550 High Output Kit v2.5 (300 Cycles). Data was processed and analyzed by the TruSight Oncology 500 Local App version 2.11.3, followed by an in-house pipeline using a second variant caller (Mutect2) and ANNOVAR for annotation of the alterations [11, 54]. For DNA analysis, single nucleotide variants, insertions/deletions, copy number variations, total mutation burden (TMB), and microsatellite instability (MSI) were calculated. For RNA analysis, putative gene fusion of around 50 fusion driver genes and RNA splice variants from EGFR, AR, or MET (e.g., MET exon 14 skipping) were explored. TMB was calculated by dividing the total number of somatic single nucleotide variants and insertions/deletions by the length of the captured coding region (~ 1.24 Mb). MSI quantitative score was calculated by interrogating 130 homopolymer MSI marker sites and defined as the proportion of MSI unstable sites to the total MSI sites. Variants were checked for germline or somatic origin using the COSMIC (catalog of somatic mutations in cancer) database [17], dbSNP, and the gnomAD database [25]. Interpretation of variants was performed using OncoKb, Varsome, and CKB [9, 31, 43].

For a systematic review of RenNETs, we screened 204 articles written in English, Japanese, or German (14 articles in other languages reporting 18 cases were excluded) using the PubMed Keywords (renal[Title] AND (neuroendocrine tumor[Title])) OR (kidney[Title] AND (neuroendocrine tumor[Title])) OR (kidney[Title] AND (carcinoid[Title])) OR (renal[Title] AND (carcinoid[Title])). 61/204 articles dealing with non-renal NETs and 4 review articles were excluded. The remaining 139 articles included 122 case reports and 17 series-based articles with a total of 227 cases. Data on sex, age, site, location, other renal disease, hormonal syndromes, outcome, Ki67 index, WHO grade, and immunohistochemical and genetic features were extracted. Outcome data such as progression-free survival (PFS) were defined as duration between the initial diagnosis and the first tumor progression, tumor-related death, or last observation. Outcome data were available in 120 cases with a mean follow-up duration of 33 months.

Thirteen RenNETs were identified (5 in-house, 8 consultations). Table 1 summarizes the most important clinical data. The mean age of the patients was 42 years (range 27–63). Metastasis was detected in 67% (8/12, data missing in 1 case) of the patients at the time of the first diagnosis. 62% (8/13) of the tumors were found in the right kidney, 38% in the left. None of the tumors was associated with horseshoe kidney. Two patients (17%) presented with Cushing syndrome; the other 11 patients (83%) had no hormone-related syndromes. No patients have multiple endocrine neoplasm type 1 (MEN1) or other hereditary/genetic tumor syndromes. The non-functioning tumors were significantly larger in size than the Cushing syndrome-associated tumors (mean 8 cm vs. 4 cm).

Three NETs were graded as a G1 (23%), 7 as G2 (54%), and 3 as G3 (23%). The mean Ki67 index of the tumors was 9% (range 2 to 33). The WHO grade was not associated with sex, age, size, hormone-related symptoms, metastases, or patients’ outcome (Table 1).

Grossly, 54% (7/13) of RenNETs were solid with a red-brown to yellow–brown cut surface (Fig. 1A) and the remaining tumors (46%) were partly cystic (Fig. 1B). Histologically, all non-functioning RenNETs had a distinct reticulated trabecular pattern (Fig. 2A, Table 1) with cubic cells slightly diastase-resistant PAS positive (Fig. 2B), while all the Cushing syndrome-associated RenNETs showed a solid growth pattern with relatively large eosinophilic and granular cells (Fig. 2C). The nuclei of these cells shaped irregular and displayed round macronucleolus and occasionally cytoplasmic eosinophilic inclusion bodies (Fig. 2D).

Immunohistochemically, all tumors were diffusely positive for CK18 and synaptophysin, while chromogranin A was diffusely expressed in both Cushing syndrome-associated RenNETs and in 3/11 non-functioning RenNETs (Fig. 3A, B). INSM1 was diffusely positive in all cases including those with a patchy chromogranin A staining. All non-functioning RenNETs expressed at least one of the pancreatic hormones (6 monohormonal, 5 multihormonal) with somatostatin in 91% (Fig. 3C), followed by PP in 63% and glucagon in 54% of the cases. Insulin and ACTH were negative in all non-functioning tumors. Both Cushing syndrome-associated RenNETs expressed diffusely ACTH (Fig. 3D), while all other hormones were negative except for a patchy somatostatin expression in one case. All non-functioning RenNETs expressed ISL1 (Fig. 3E), while Cushing-associated tumors were negative (Fig. 3F). SATB2 was diffusely positive in all non-functioning tumors and negative in Cushing syndrome cases. CDX2, TTF1, and PAX8 were consistently negative. Membranous SST2 labeling was found in 49% of non-functioning RenNETs but in none of the Cushing syndrome-associated RenNETs (Table 1).

NGS was successfully performed in all tumors. The known NET-related gene alterations such as ATRX, DAXX, MEN1, or TSC1/2 or NEC-related genes such as TP53, RB1, or PIK3CA-related genes were not detected in any of the cases. No gene with a pathogenic variant (class 5) was identified. Gene alterations with probably pathogenic variations (class 4) were found in three non-functioning tumors (23%) affecting SDHA (case 2), ARID1B (case 6), and ASXL1 (case 9).

Gene alterations of class 3–T4 (variant of unclear significance with probably pathogenic tendency) were found in four non-functioning tumors affecting PMS2 (case 11), NUP93 (case 4), TOP2A (case 2), and SNCAIP (case 3). Other gene changes with unclear pathogenic significance (class 3) are listed in Supplementary Table 2. Fusion genes were not detected. The median value of TMB was 0.8 (range 0–5.5). The median MSI quantitative score was 1.72% (range 0–5.88). None of the cases showed a high MSI score (> 10). The tumor with the highest MSI score (5.88 in case 11) showed retained immunohistochemical staining for mismatch repair proteins (MLH1, PMS2, MSH2, MSH6). Loss of SDHB expression was not observed in two cases (case 9 and case 13) with class 2 SDHD mutations (Supplementary Table 2).

The clinical data of 227 RenNETs extracted from published articles is summarized in Table 2. Of the 194 cases in which hormonal status was documented, 15 (8%) had hormonal symptoms, of which Cushing syndrome was being the most frequent (7/15), followed by carcinoid syndrome (5/15). The remaining three RenNETs with hormonal syndrome included an insulin-secreting tumor with a hypoglycemic syndrome [45], a glucagon-producing tumor with a glucagonoma-like syndrome [18], and a tumor with a carcinoid-like syndrome [24]. Follow-up data were available in 120 cases (mean 33 months). Large tumor size (6 cm or larger, p = 0.01) and presence of metastasis at the time of diagnosis (p < 0.001) were significantly associated with poor patient outcome based on PFS, while age, sex, hormone-related syndrome, and horseshoe kidney were not. Data on WHO grade, available in only 58 cases, and Ki67 index, available in 41 cases, were not associated with outcome. Pancreatic hormone expression was reported to be positive in 60% (6/10) for glucagon, in 50% (7/14) for somatostatin, in 25% (3/12) for PP, and in 59% (7/12) for serotonin. The transcription factor ISL1 was examined in one case and was positive [12], while TTF1 and PAX8 were negative in all examined cases (30 and 27 cases, respectively). Molecular analysis was performed in four studies [15, 34, 44, 53]. Loss of heterozygosity (LOH) on 3p21 was reported in 5 of 11 cases [15, 34, 44]. NGS analysis was performed in 9 RenNETs and revealed variable mutation profiles [44]. The gene abnormalities which were most frequently found included mutations of CDH1 and TET2, with three mutations in two cases. Next in frequency were LOH 3p and mutations in AKT3, ROS1, PIK3P2, BCR, and MYC [44].