Evaluation of an institutional series of low-grade oncocytic tumor (LOT) of the kidney and review of the mutational landscape of LOT

The 2022 WHO classification of urinary and male genital tumors introduced several novel kidney entities exhibiting eosinophilic/oncocytic features with specific mutational backgrounds. Thus, molecular techniques, such as next-generation sequencing (NGS), became more commonly used for their evaluation. We studied 12 low-grade oncocytic tumors (LOT) of the kidney (from 11 patients), identified in a cohort of 210 eosinophilic/oncocytic renal tumors, diagnosed in our institution between October 2019 and May 2023, which represented 5.7% (12/210) of all eosinophilic/oncocytic renal tumors during this period. We reviewed their clinicopathologic, histologic, and immunohistochemical features, as well as their mutational profiles. We also reviewed the literature on NGS-derived data of LOT, by selecting papers in which LOT diagnosis was rendered according to the criteria proposed initially. Median age was 65 years (mean: 63.5; range 43–79) and median tumor size was 2.0 cm (mean: 2.2; range: 0.9–3.1). All tumors were positive for PAX8, CK7, and GATA3, and negative or focally positive for CD117/KIT. We found the following gene mutations: MTOR ((6/11), 54.5%)), TSC1 ((2/11), 18.2%)), and 1 had both NOTCH1 and NOTCH4 ((1/11), 9.1%)). Wild-type status was found in 2/11 (18.2%) patients and one tumor was not analyzable. A review of 8 previous studies that included 79 LOTs revealed frequent mutations in the genes that regulate the mammalian target of rapamycin (mTOR) pathway: MTOR (32/79 (40.5%)), TSC1 (21/79 (26.6%)), and TSC2 (9/79 (11.4%)). Other mutated genes included PIK3CA, NF2, and PTEN, not typically known to affect the mTOR pathway, but potentially acting as upstream and downstream effectors. Our study shows that LOT is increasingly diagnosed in routine practice when applying the appropriate diagnostic criteria. We also confirm that the mTOR pathway is strongly implicated in the pathogenesis of this tumor mainly through MTOR, TCS1, and TSC2 mutations, but other genes could also be involved in the pathway activation, especially in LOTs without “canonical” mutations. Supplementary Information The online version contains supplementary material available at 10.1007/s00428-023-03673-9.

In this study, we aimed to interrogate an institutional case series of LOT with two in-house-developed multi-gene NGS panels.We also performed a literature review to clarify the genetic landscape of LOT, focusing exclusively on NGS data and including only tumors diagnosed according to the original description of LOT by Trpkov and Hes [3,4].
All clinicopathologic investigations were conducted according to the principles of the Declaration of Helsinki and all information regarding the human material used in this study has been managed using anonymous numerical codes.The study has been approved by the Review Board of the Area Vasta Emilia Centro-AVEC (IRB approval 3386/2018protocol).All patients included in the study provided informed consent, after a consultation with the investigators.

Molecular genetic analysis using a custom-designed NGS panel
DNA from formalin-fixed paraffin-embedded blocks was extracted using two to four 10-μm sections, under microscopic guidance from the representative tumor areas, identified by a pathologist on H&E slide.Extracted DNA was used for amplicon library preparation using two laboratory-developed multi-gene panels.The first panel (panel 1) allowed amplifying a total of 623 amplicons (69.7 kb, human reference sequence hg19/GRCh37) of the following genes: AKT (whole coding sequence (CDS)), BRAF (exons 11, 15), EP300 (CDS), HRAS (exons 2-4), KRAS (exons 2-4), MEN1 (CDS), MTOR (CDS), NOTCH1 (CDS), NOTCH2 (CDS), NOTCH3 (CDS), NOTCH4 (CDS), and NRAS (exons 2-4).The second panel (panel 2) allowed amplifying a total of 280 amplicons (29.24 kb, human reference sequence hg19/GRCh37) in the whole CDS of the following genes: FH, FOXL2, HMGA1, MED12, TSC1, and TSC2.Briefly, about 30 ng of input DNA was used for NGS library preparation for each panel using the AmpliSeq Plus Library Kit 2.0 (Thermo Fisher Scientific, Waltham, MA, USA).Templates were then sequenced using an Ion 530 chip and the results were analyzed with the IonReporter tools (version 5.18, Thermo Fisher Scientific) and GenomeBrowser Tool (https:// www.golde nhelix.com/).According to the previously reported validation, only mutations present in at least 5% of the total number of analyzed reads and observed in both strands were considered as mutational calls [31].The Varsome tool (https:// varso me.com/, updated to April 2023) was used to evaluate the American College of Medical Genetics and Genomics classification of each mutation [32].

Literature review of mutational landscape of LOTselection criteria
We reviewed the literature on the mutational profile of LOT to select only studies encompassing NGS data and published by April 2023, in line with the study time frame and selection criteria.We only selected studies in which the diagnosis of LOT had been rendered using the histomorphologic and immunohistochemical criteria proposed by Trpkov and Hes [3,4].In our review, we excluded tumors with absent and/ or incomplete clinicopathologic and immunohistochemical data (CK7, CD117/KIT), in which a definitive diagnosis of LOT could not be established, using the accepted criteria [19,[33][34][35][36][37][38][39][40].Indeed, many LOTs have likely been reported in the literature and in The Cancer Genome Atlas (TCGA) as other entities (for example, as oncocytomas and ChRCCeosinophilic) and we did not include such tumors because a definitive diagnosis of LOT could not be established [19,[33][34][35][36][37][38][39][40].

Molecular results
Of the 12 samples analyzed by NGS (Fig. 2), one was not evaluable due to low-quality DNA.Of the remaining 11 specimens, 6/11 (54.5%) harbored MTOR mutations (5 pathogenic and 1 likely pathogenic variants) and 2/11 (18.2%)TSC1 mutations (with identical likely pathogenic variant); of note, all mutations were found in different specimens with no overlap (Fig. 2).Notably, two LOTs (tumors #10 and #11) with identical TSC1 mutations were obtained from one patient; analysis of the non-neoplastic tissue of this patient had the same TSC1 substitution, which strongly suggested a diagnosis of TSC (Table 1 and Fig. 2).This patient underwent clinical-genetic examination which confirmed the diagnosis of TSC; the patient had epileptic episodes due to multiple cortical tubers and had skin hamartomas.Of note, in 1/11 (9.1%) tumors, there were two co-occurring mutations: NOTCH1 mutation (p.Glu515Lys) and NOTCH4 mutation (p.Asp272Gly).While in December 2022 these variants were classified as variants of unknown significance (VUS) in the Varsome database, in June 2023, both were reclassified as likely benign.

Literature review of the mutational landscape of LOT
Using the outlined criteria, 8 previous studies were identified for review, resulting in a total of 79 LOTs tested by NGS panels, including tumors from the current study (summarized in Tables 2 and 3) [20-25, 28, 29].The most frequently detected mutations were those typically known to affect the mTOR pathway: MTOR (32/79 (40.5%)),TSC1 (21/79 (26.6%)), and TSC2 (9/79 (11.4%)) (Tables 2 and 3).Two tumors (2.5%) showed two different TSC1 mutations, and 5 tumors (6.3%) had co-mutations in two of these three genes (2 tumors with MTOR and TSC2 mutations, and 3 tumors with TSC1 and TSC2 mutations).However, other genes, "not typically known" to affect the mTOR pathway, such as STK11, PTEN, FOXP1, FGFR3, NF2, MET, PIK3CA, RHEB, CDKN2A, EZH2, SETD2, and PIK3CA, were also mutated in these tumors (Table 2).These genes may be potentially involved in the abnormal activation acting as upstream and/or downstream effectors of the mTOR pathway.Of note, 12/24 (50%) of these alternative mutations were not concomitant with MTOR, TSC1, and TSC2 mutations.Notably, PIK3CA was not concomitant and was found as pathogenic in 5/6 (83.3%) tumors, which may suggest that PIK3CA mutation may be "sufficient" on its own to activate the mTOR pathway, without the concurrent MTOR, TSC1, and TSC2 mutations.Overall, 14/24 (58.3%) of these alternative mutations were found as pathogenic and/or likely pathogenic (see Tables 2 and 3).There were 9/79 (11.4%) tumors with a wild-type (WT) status and without detectable mutations by the NGS panels used.However, due to variability and scope of the NGS panels used in different studies, we cannot rule out that other relevant undetected genes may have been implicated in these tumors.

Discussion
One of the more challenging areas in uropathology is the differential diagnosis of renal tumors with eosinophilic/ oncocytic features [1][2][3][4][5][6][7].An increasing number of studies included molecular evaluations that facilitated the recognition of several new entities, which were separated from the spectrum of tumors previously incorrectly labelled as oncocytoma, eosinophilic ChRCC, or considered "unclassified oncocytic tumors/RCCs" [1][2][3][4][5][6][7].In recent years, mTOR pathway prompted increasing interest and focus of study as an underlying theme for several novel and emerging entities [1-7, 14-30, 33, 35-37].mTOR was known to be involved in the pathogenesis of angiomyolipoma (both in syndromic and non-syndromic scenarios), but mutations in genes that regulate this pathway (TSC1, TSC2, and MTOR) were also rarely found in some common renal tumors, such as ChRCC, clear cell RCC, and papillary RCC [1][2][3][4][5][6][7][8][9][37][38][39].Nevertheless, the mTOR pathway was found more recently to be primarily involved in the pathogenesis of specific "pink tumors", such as ESC RCC, EVT, LOT, and RCC FMS [1-7, 14-30, 33].The shared molecular alterations affecting the mTOR pathway led some authors to wonder if these tumors should be grouped as "TSC/MTOR-associated renal tumors", although they demonstrated different clinicopathologic, histologic, and immunohistochemical features [25][26][27].However, the presence of distinct morphologic and immunohistochemical features, as well as the differences in gene involvement in different entities (for example, ESC RCC showed almost exclusively bi-allelic somatic TSC2 mutations) led some authors to conclude that although these neoplasms share mTOR pathway molecular mechanisms, they should be considered as distinct entities [25][26][27].In the present study, we evaluated a case series of LOT of the kidney with laboratory-developed multi-gene NGS panel.We intentionally selected a time frame following the original description of LOT to obtain more reliable data on the prevalence of this tumor in routine diagnostic practice [3,4].We identified 12 LOTs, which represented 0.7% (12/1670) of the total renal tumor volume and 5.7% (12/210) of all oncocytic/eosinophilic renal tumors.These frequencies are higher than those reported in previous studies and suggest that the application and familiarity with well-defined histologic and immunohistochemical criteria increases the number of LOT cases diagnosed in routine practice [17,25,27].LOT shows an oncocytoma-like appearance with a prominent solid architecture, absence of peripheral capsule, and rare entrapped renal tubules at the periphery.At high-power magnification (B, H&E, original magnification 100 ×), LOT displays oncocytic/eosinophilic cells with low-grade nucleoli, absence of irregular/raisonoid nuclear appearance, and no significant perinuclear halos.In the central area, LOT exhibits the so-called boats in a bay arrangement with hypocellular areas occupied by loosely arranged tumor cells (C, H&E, original magnification 100 ×).At immunohistochemistry, LOT results positive for GATA3 (D, original magnification 100 ×) and CK7 (E, original magnification 100 ×), but negative for CD117/KIT (F, original magnification 100 ×).The chromogen (labeling) for GATA3, CK7, and CD117/c-kit is DAB We also confirm a consistent and uniform GATA3 reactivity in all LOTs included in the current study, which supports its usefulness as an additional marker in the diagnosis of LOT [29,40].In our experience, GATA3 can be also variably and focally expressed in ChRCC, but it is essentially negative in all other eosinophilic/oncocytic renal tumors.Based on the 8 reviewed studies and an aggregate cohort of 79 LOTs, the majority (57/79 (72.1%)) exhibited at least one mutation in Fig. 2 Genomic profile of the LOTs in the cohort.These two specimens (indicated by asterisk symbols) were obtained from one patient and the analysis on the non-neoplastic tissue revealed the same TSC1 substitution.At December 2022, these variants (indicated by upward-pointing arrowheads) were classified as VUS in the Varsome database, but in June 2023, both were reclassified as likely benign Table 2 Mutational landscape of LOTs (our case series and review of the literature) Eight previous studies [20-25, 28, 29] and a total of 79 LOTs (by adding our case series) were included in this review (see "Case series" in "Materials and methods" and "Results")   RHEB, CDKN2A, EZH2, SETD2, NOTCH1, and NOTCH4).
Although these genes are known to be involved in distinct and well-defined intracellular mechanisms, they could also display promiscuous and heterogenous effects, potentially causing the up-and down-stream activation of the mTOR pathway (Fig. 3) [24,29,[41][42][43].Although it is difficult to establish specifically which of these mutations affect the mTOR pathway or are "simply passenger mutations", 14/24 (58.3%) were found as pathogenic and/or likely pathogenic (Table 3).Moreover, 12/24 (50%) were not concomitant with the mutations in MTOR, TSC1, and TSC2 (Table 2); among these genes, PIK3CA was pathogenic/likely pathogenic and not concomitant in 5/6 (83.3%) tumors.Kapur et al. analyzed the relationship between the detected mutations, the mTORC1 structure, and the intracellular levels of p-S6 and p-4EBP1 (specific markers of the mTORC1 activation) to evaluate the impact of the individual mutations on mTORC1 activity [21].They found that levels of mTORC1 activation varied depending on the type of mutation (for example, mTORC1 activation was lower in tumors with MTOR mutations) [21].These data, along with our findings, suggest that some of these "non-canonical" mTOR-activating mutations may be "sufficient" on their own to activate the mTOR pathway.A subset of mutations could also cooperate with "canonical" mTOR-activating mutations in determining and increasing the final degree of mTORC1 function (e.g.tumor #2 in study by Mohanty et al. with STL11 and TSC1 mutations [24]).These results suggest that a subset of LOTs and especially tumors without detectable MTOR, TSC1, and TSC2 mutations may have other mutated genes that may potentially activate the mTOR pathway.
To conclude, our results show that LOT is a straightforward diagnosis in routine practice, typically requiring morphologic and limited immunohistochemistry evaluation, as outlined in the initial studies and in the current WHO classification [1][2][3][4].This institutional LOT cohort assembled after the initial publications indicates that LOT appears to be more frequent than previously reported.Our review of the molecular landscape of LOT shows that the mTOR pathway is strongly implicated in the pathogenesis of LOT.In tumors with no mutations in TCS1, TSC2, and MTOR genes, mutations in other genes may potentially affect the mTOR pathway and additional studies utilizing broader and standardized molecular gene panels may further clarify this issue.

Table 1
Clinicopathologic features and NGS results of the analyzed cohort * mutation was detected in the non-neoplastic tissue, and the subsequent clinical-genetic examination confirmed a diagnosis of TSC NGS, next-generation sequencing; VAF, variant allele frequency; ACMG, American College of Medical Genetics and Genomics; F, female; M, male; L, left; R, right; NA, not available; NED, not evidence of disease; WT, wild-type status; P, pathogenic; LP, likely pathogenic; LB, likely benign; VUS, variant of unknown significance; SNV, single-nucleotide variant; INDEL, insertiondeletion; LOT/LOTs, low-grade oncocytic tumor/tumors; TSC, tuberous sclerosis complex (syndrome) Patient number

Table 3
Mutational landscape of LOTs (our case series and review of the literature)Eight previous studies[20-25, 28, 29]with a total of 79 LOTs (including current study) were included in the review LOT, low-grade oncocytic tumor; WT, wild-type status; ACMG, American College of Medical Genetics and Genomics; P, pathogenic; LP, likely pathogenic; NS, not specified; LB, likely benign; B, benign; VUS, variant of unknown significance