L1CAM Expression and Molecular Alterations Distinguish Low Grade Oncocytic Tumor (LOT) from Eosinophilic Chromophobe Renal Cell Carcinoma

Abstract

Renal low-grade oncocytic tumor (LOT) is a recently recognized renal cell neoplasm designated within the “other oncocytic tumors” category in the 2022 WHO Classification. While the clinicopathologic, immunohistochemical, and molecular features reported for LOT have been largely consistent, the data is relatively limited. The morphologic overlap between LOT and other low grade oncocytic neoplasms, particularly eosinophilic chromophobe renal cell carcinoma (E-chRCC), remains a controversial area in renal tumor classification. To address this uncertainty, we characterized and compared large cohorts of LOT (n=67) and E-chRCC (n=69) and revealed notable differences between the two entities. Clinically, LOT predominantly affected females whereas E-chRCC showed a male predilection. Histologically, while almost all LOT dominated by a small-nested pattern, E-chRCC mainly showed solid and tubular architectures. Molecular analysis revealed that 87% of LOT cases harbored mutations in the TSC-mTOR Complex 1 (mTORC1) pathway, most frequently in MTOR and RHEB genes; a subset of LOT cases had chromosomal 7 and 19q gains. In contrast, E-chRCC lacked mTORC1 mutations and 60% of cases displayed chromosomal losses characteristic of chRCC. We also explored the cell of origin for LOT and identified L1CAM, a collecting duct and connecting tubule principal cell marker, as a highly sensitive and specific ancillary test for differentiating LOT from E-chRCC. This distinctive L1CAM immunohistochemical labeling suggests the principal cells as the cell of origin for LOT, unlike the intercalated cell origin of E-chRCC and oncocytoma. The ultrastructural analysis of LOT showed normal-appearing mitochondria and intracytoplasmic lumina with microvilli, different from what has been described for chRCC. Our study further supports LOT as a unique entity with benign clinical course. Based on the likely cell of origin and its clinicopathologic characteristics, we propose that changing the nomenclature of LOT to “Oncocytic Principal Cell Adenoma of the Kidney” may be a better way to define and describe this entity.

INTRODUCTION

The differential diagnosis of renal neoplasms with eosinophilic or oncocytic cytoplasm, particularly low-grade oncocytic renal neoplasms, has undergone significant evolution in the last decade. One important development is the emergence of a heterogeneous group of tumors with overlapping, yet somewhat distinctive, morphological features when compared to classic renal oncocytoma and eosinophilic chromophobe renal cell carcinoma (E-ChRCC). The diagnostic challenges prompted by these tumors have led to the inclusion of a novel “other oncocytic tumors of the kidney” category in the current WHO classification system (5th edition).^{1, 2} However, recognition of emerging entities within this category of oncocytic tumors has also generated controversy regarding diagnostic criteria and work-up of tumors in this morphologic spectrum,^3–5 highlighting the need to integrate new knowledge gained from molecular analyses with traditional morphologic criteria and ancillary tools.

Low-grade oncocytic tumor (LOT) is one of the emerging and distinct oncocytic renal neoplasms. It typically consists of densely packed small nests of oncocytic cells with low-grade, round to oval nuclei that often display perinuclear halos and slight nuclear membrane irregularities but not the prominent nuclear irregularities (“raisinoid” features) commonly found in E-chRCC. Immunohistochemically, most reported cases of LOT consistently exhibit strong and diffuse immunoreactivity to CK7 and are negative for CD117.^6–20 Molecularly, these cases have been found to harbor TSC/mTOR pathway alterations such as recurrent driver mutations in MTOR, TSC1, and RHEB in recent studies.^{7, 11, 13, 15–19} Although most reported cases have not been syndrome-associated, three reports identified an association with Tuberous Sclerosis Complex (TSC) in a total of 5 patients.^{12, 13, 16} Chromosomal copy number alterations have also been reported, however, the changes appear to be variable among analyzed tumors.^{6, 12}

E-chRCC is a morphologic variant of chromophobe renal cell carcinoma (chRCC) that was first described by Thoenes et al.²¹ and has been used to describe tumors predominantly composed of eosinophilic cells and demonstrating irregular nuclei and perinuclear halos.¹ At the molecular level, while a large majority of classic chRCC show loss of chromosomes 1, 2, 6, 10, 13, 17, and 21,^{22, 23} E-chRCC has been found to exhibit fewer or no genetic losses; however, this finding is not consistent among studies, possibly reflecting challenges in applying consistent diagnostic criteria for E-chRCC.^23–25 Overall, chRCC displays a low rate of somatic mutations, with TP53 and PTEN identified as the only two significantly mutated genes.²³

Given the morphologic overlap and uncertain relationship between LOT and E-chRCC, we aimed to characterize and compare large cohorts of LOT and E-chRCC to better delineate their clinicopathologic, immunohistochemical, and molecular features. We also explored the cellular origin of LOT and identified a novel immunohistochemical marker that is highly sensitive and specific for differentiating LOT from E-chRCC. To our knowledge, the current study analyzed the largest cohort of LOT to date and represents the first study to compare large cohorts of LOT and E-chRCC with comprehensive characterizations and long-term clinical follow-up data.

MATERIAL AND METHODS

Case selection

Following Institutional Review Board (IRB) approval, a pathology database search was performed to identify consecutive in-house partial/radical nephrectomy cases with a rendered diagnosis of E-chRCC, low-grade oncocytic unclassified renal cell carcinoma, or LOT between the years 2000 and 2022 at Memorial Sloan Kettering Cancer Center. Among 275 cases that were identified, 272 had material available for study. Histologic review was performed for all cases with available materials (glass or scanned slides); additional immunohistochemical analyses were performed if the original diagnostic work-up was incomplete (e.g. CK7 and CD117 for cases morphologically consistent with LOT or E-chRCC; SDHB, FH/2SC, TFEB/TFE3, etc. for cases with morphologic features suspicious for these relatively newly recognized and molecularly defined entities). Based on the histologic re-review and additional analysis, and using current diagnostic criteria, we identified 3 groups: LOT (n=51), E-chRCC (n=69), and others (n=152). To analyze the characteristics of LOT more comprehensively, we included 16 additional LOT tumors: 11 consultation cases submitted by outside hospitals/pathologists, and 5 cases from the University of Pennsylvania with ultrastructural analysis (approved by the respective IRB).

Clinicopathologic analysis

Clinical and follow-up information was collected from electronic medical records documenting age, sex, length of follow-up, and vital status at the last follow-up. Gross pathology descriptions were used to collect tumor size, laterality, and focality. Pathological tumor stage (pT) was assigned according to the eighth edition of the American Joint Committee on Cancer (AJCC) cancer staging manual. Histomorphologic features were reviewed, and the following parameters were recorded: dominant tumor architecture, presence/absence of sharply demarcated areas of stromal edema, perinuclear clearing/halos, irregular nuclear contour, and any other unusual features. The 3 dominant architectural patterns used for analysis included 1) “small nested” (small solid nests, oncocytoma-like), 2) “classic chRCC” (solid, sheets-like or broad trabecular with incomplete vascular septa), and 3) “tubular” (tubular, alveolar, or tubulocystic growth).

Immunohistochemical analysis

Immunohistochemistry (IHC) was performed on 5-μm-thick sections from formalin-fixed, paraffin-embedded tissue utilizing antibodies for CK7 (OV-TL 12/30, ThermoFisher; 1:4000), CD117 (Cat# A4502, Dako; 1:500), and two antibodies downstream of the mTOR pathway: Phospho-S6 Ribosomal Protein (Ser240/244) (p-S6) (D68F8, Cell signaling; 1:2000) and Phospho-4E-BP1 (Thr37/46) (p-4EBP1) (236B4, cell signaling; 1:200), as well as L1CAM (14.10, Biolegend,1:400). CK7 staining was scored as: diffusely positive, focal/patchy, or negative (0–5% scattered rare cells); CD117 was scored as positive, focal/patchy, and negative. L1CAM was scored as diffusely positive, focal/patchy, and negative (0–5% scattered rare cells). Immunostaining scores (H-scores) for p-S6 and p-4EBP1 were determined as [H= intensity (0–3) x percentage of positive cells (1–100)].

Molecular analysis

Thirty-seven cases representing the morphologic spectrum of the study cohorts were selected for evaluation utilizing a next-generation sequencing (NGS) platform with 505 cancer-related genes (MSK-IMPACT).²⁶ These included LOT (n=23), E-chRCC (n=11), and 3 cases of oncocytic neoplasm NOS that showed equivocal histologic features for E-chRCC or LOT. DNA was extracted from the macro-dissected formalin-fixed paraffin-embedded (FFPE) tissue blocks of both the tumor and normal kidney. Somatic mutations were called after germline single-nucleotide variants (SNVs) detected in the paired normal sample were filtered out. The functional impact of detected mutations was classified using OncoKB (http://oncokb.org), a precision oncology database maintained at MSKCC.²⁷ Variants not annotated in OncoKB were further assessed in the COSMIC database²⁸ and by literature review. Copy number analysis was conducted using coverage-based analysis and FACETS (v0.5.6), an algorithm incorporating allele-specific copy number assessment and tumor purity estimation for detecting copy number alterations and loss of heterozygosity.²⁹ Arm-level copy number alterations (including loss of heterozygosity) were defined as copy-number changes occurring in a segment that comprised at least 50% of the chromosomal arm with estimated tumor cell fraction at 20% or higher; copy number alterations with affected tumor cell fraction less than 20% were classified as subclonal events.

Electron Microscopy (EM)

Transmission electron microscopy (TEM) was used to examine the ultrastructure of five LOT tumors and the analysis was conducted at the University of Pennsylvania. The tissue was either fixed in EM fixative, consisting of 2% glutaraldehyde and 4%formaldehyde in 0.1M sodium cacodylate buffer, or reprocessed from formalin-fixed paraffin-embedded (FFPE) tissue for TEM. Following a wash with sodium cacodylate buffer, the samples were post-fixed with 1 % osmium tetroxide for 1 hour, dehydrated in graded concentrations of alcohol, infiltrated in a mix of propylene oxide and epoxy resin (epon), and embedded in epon overnight. FFPE tissue was first deparaffinized, followed by routine processing. Semi-thin sections were cut at 2 microns using a PowerTome XL ultramicrotome and stained with 1% toluidine blue. Ultrathin 90 nm sections were placed on copper grids, stained with 20% uranyl acetate and 0.4% lead citrate, and examined with a Hitachi H-7000 transmission electron microscope operated at 80 kV. Digital images were acquired using an Advanced Microscopy Techniques camera system.

RESULTS

Among a total of 288 nephrectomies of renal oncocytic/eosinophilic neoplasms re-reviewed, using diagnostic criteria of the 2022 WHO classification system,¹ we identified 67 LOT (Fig. 1) and 69 E-chRCC (Fig. 2). The remaining 152 cases exhibited histologic and/or IHC features not fitting with either entity, most of which would be considered renal oncocytic neoplasm of low malignant potential, NOS.¹ For this study, we focused on characterizing and comparing the first two groups, LOT and E-chRCC. The clinicopathologic features of these two groups are summarized in Table 1 and detailed in Fig. 3 and Supplementary Table 1.

Figure 1.

Low-grade oncocytic tumor (LOT). A) Representative gross appearance is a well-demarcated tan-yellow mass with focal hemorrhage or cystic change. B-C) Tumor is mostly solid with focal abrupt transition to stromal edema or hemorrhage with tumor cells floating as cords or trabeculae in the edematous background. D) Tumor cells with densely eosinophilic cytoplasm, perinuclear clearing, and smooth nuclear membranes. E-F) Some cases show vacuolated cytoplasm (E) or focal tubulocystic patterns (F). G) Osseous metaplasia and associated extramedullary hematopoiesis. H) CK7 immunostain. I) CD117 immunostain.

Figure 2.

Eosinophilic chromophobe renal cell carcinoma (chRCC). A) Gross image shows a large tumor with tan-yellow cut surfaces and nodular appearance with fibrous septa. B-C) Two examples of the “classic chRCC” architecture - solid sheets or broad trabecular arrangements with incomplete vascular septa. D) Tubular architecture. E) Small nested architectural pattern, perinuclear halos, and irregular nuclear contour are present. F) CD117 shows moderate membranous staining. G-I) CK7 immunostain can be diffuse (G), focal/patchy (H), or negative (I).

Table 1.

Summary of clinicopathologic features in LOT and E-chRCC cohorts.

	LOT (n=67)	E-chRCC (n=69)
# of patients	65	68
Age	65 (36–89)	61 (29–80)
Sex
Male	20 (31%)	47 (69%)
Female	45 (69%)	21 (31%)
Tumor size (cm) * §	3.85 (1–12)	3.25 (0.9–20)
Pathologic stage §
pT1	56 (88%)	60 (87%)
pT2	7 (10%)	4 (6%)
pT3	1 (1.7%)	5 (7%)
Nx/N0	64 (100%)	69 (100%)
Mx/M0	64 (100%)	6 (100%)
Dominant architecture
Small nested	67 (100%)	10 (14%)
Classic chRCC#	0	33 (48%)
Tubular	0	26 (38%)
Perinuclear halos
Absent/rare cells	8 (12%)	10 (14%)
Present (patchy/diffuse)	59 (88%)	59 (86%)
Irregular nuclear contour
Absent	55 (82%)	1 (1%)
Present (focal/diffuse)	12 (18%)	68 (99%)
Immunohistochemistry
CK7 – diffusely positive	59/60 (98%)	18/50 (36%)
CD117 – positive	1/60 (2%)	47/50 (94%)
L1CAM – diffusely positive	32/32 (100%)	0/37 (0%)
GATA3 – positive (any)	23/23 (100%)	8/23 (35%)
Follow-up (year) *	5.8 (0.2–19)	7.0 (0.2–21)
Recurrence/metastasis	0	2 (3%)

^*Median (range)

^§Missing data in 3 outside cases in the LOT cohort

^#Classic chRCC architecture: solid, sheet-like or broad trabecular arrangement with incomplete vascular septa

Figure 3.

Detailed clinicopathological characteristics of LOT and E-chRCC cohorts. Each bar represents a case. “*” marks 3 metachronous tumors from a patient in the LOT cohort, “^#” marks 2 metachronous tumors from a patient in the E-chRCC cohort.

Clinicopathologic features: LOT cohort

The LOT cohort consisted of 67 cases from 65 patients. The median age of patients was 65 years (36–89) at diagnosis and there was a female predominance (M:F = 1:2.3). The most common initial reported diagnosis for these cases was E-chRCC (n=44; 66%), followed by low grade unclassified RCC (n=14; 21%); 8 recent cases were reported as LOT (12%). One patient had multiple, bilateral tumors resected at three different time points spanning 11 years, and 3 metachronous LOT tumors were included in the analysis. The frequency of LOT among all nephrectomies performed at our institution within the study period (2000–2022) is approximately 0.6%. Grossly, the tumors were well-demarcated and had tan-yellow or brown cut surfaces; scattered areas of hemorrhage or focal cystic changes were seen in a subset of cases (Fig. 1A). Microscopically, all cases displayed compact, small nests arranged in a predominantly solid growth pattern; most cases (53/67, 79%) contained focal areas showing an abrupt transition from the solid small nests to stromal edema or hemorrhage, with tumor cells floating in a cord-like, trabecular, and tubular fashion (Fig. 1B–C). The neoplastic cells mostly exhibited densely eosinophilic cytoplasm with perinuclear clearing/halos, and round to oval nuclei with smooth nuclear membranes or subtle nuclear membrane irregularity (Fig. 1D). Subsets of cases showed variation in the cytoplasmic quality and architecture such as granular/reticular or vacuolated cytoplasm, and focal cystic or tubulocystic patterns (Fig. 1E–F). One case showed osseous metaplasia and associated extramedullary hematopoiesis (Fig. 1G).

By IHC, 59 of 60 (98%) LOT cases with tissue available showed diffuse positivity for CK7, and 1 case (T16) showed only focal staining. Except for one case (T22) showing focal CD117 staining, all other cases were negative for CD117 (Fig. 1H–I and Supplementary Fig. 1). All cases were confined within the kidney (pT1-T2) except for one tumor protruding into a renal sinus segmental branch of renal vein (pT3a). There was no evidence of recurrence or metastasis among all 56 patients with follow-up data (range, 0.2–19 years; median, 5.8 years).

Clinicopathologic features: E-chRCC cohort

The E-chRCC cohort consisted of 69 cases from 68 patients with a median age of 61 years (29–80) and showed a male predominance (M:F = 2.2:1). Two metachronous E-chRCC tumors from a patient with bilateral partial nephrectomies were included. The frequency of E-chRCC among all nephrectomies within the same study period is approximately 0.85%. Macroscopically, E-chRCC also had tan-yellow or brown cut surfaces, although nodular appearance and fibrous septation were seen in a few large tumors (Fig. 2A). There were 3 main architectural patterns noted: solid sheets or broad trabecular arrangements with incomplete vascular septa (annotated as “classic chRCC”) (n=33; 48%), tubular (n=26; 38%), and small nested (n=10; 14%) (Fig. 2B–E). Consistent with the morphologic criteria used for E-chRCC in this study, 59 (86%) cases displayed diffuse or patchy presence of perinuclear halos and 68 (99%) had scattered or widely distributed cells with irregular nuclear contour; all (100%) cases exhibited granular or densely eosinophilic cytoplasm. Meanwhile, 46 of 50 (92%) tumors with IHC results showed immunoreactivity for CD117, which ranged from focal/patchy to strong diffuse membranous staining (Fig. 2F). For CK7, 28 (56%) E-chRCC tumors had either diffuse (n=18) or patchy (n=10) positivity, while the remaining 22 (44%) were negative or showed scattered positive cells (Fig. 2G–I and Fig. 3). Five E-chRCC cases were pT3a; over a median follow-up of 7 years (range, 0.5–21), one patient (0.15%) developed metastasis, and another had local recurrence.

Molecular alterations in LOT and E-chRCC

Molecular analysis by targeted sequencing was conducted for 23 LOT and 11 E-chRCC, as well as in 3 oncocytic neoplasms, NOS which were difficult to distinguish from LOT or E-chRCC (Fig. 4A and Supplementary Table 2). The LOT cohort showed a high prevalence (20/23; 87%) of somatic mutations in genes regulating the TSC/mTOR complex 1 (mTORC1) pathway, most frequently involving MTOR (n=11) and RHEB (n=7) with MTOR L2427Q/R and RHEB Y35N as the most common mutations (Fig. 4A–B). The high frequency and mutually exclusive presence of mutations in the mTORC1 pathway as well as a lack of other recurrently mutated genes in the LOT cohort strongly support that mTORC1 signaling is the main driver of tumor development. Interestingly, while none of the cases had TSC1 or TSC2 germline mutations, three patients, including one with multiple LOT tumors, harbored germline mutations of APC, LZTR1, or MUTYH, the significance of which is currently unclear (Fig. 4A). The three cases without detectable mTORC1 pathway-related alterations by our targeted NGS panel nonetheless showed typical clinicopathologic features of LOT.

Figure 4.

Molecular analysis of LOT and E-chRCC cohorts. A) Mutational landscape based on a targeted DNA sequencing panel in 23 LOT, 11 E-chRCC, and 3 oncocytic NOS cases. Mutated genes are listed on the left and denoted by individual rows. Individual tumors are presented as columns and labeled at the bottom. B) Pie chart summarizing the mutations of TSC/mTOR complex 1 pathway mutations detected in the LOT cohort. C) Copy number alterations in LOT (n=22), E-chRCC (n=10), and Oncocytic NOS (n=3) cases. WGD- whole genome doubling.

By FACET copy number analysis (Fig. 4C), arm-level gains of chromosomes 19q and 7 were identified in 12 (out of 22, 55%) and 8 (36%) LOT cases, respectively. However, the large majority of 19q gains appeared to be sub-clonal. Additionally, one case (T21) (Supplementary Fig.1) had widespread chromosomal gains consistent with whole genome doubling (WGD). The remaining 9 (41%) cases had a flat genome.

Among LOT cases with mTORC1 gene mutations, 95% (19/20) were monoallelic (MTOR, RHEB, and TSC2) alterations with a retained wildtype allele at the respective gene locus; one tumor (T21) showed likely biallelic inactivation of TSC1 due to loss of heterozygosity. Additionally, there was a trend of co-occurrence of RHEB (7q36) mutation and chromosome 7 gain (i.e., 5 of 7 RHEB mutated cases had chromosome 7 gain; Fisher Exact test p=0.05). Moreover, in line with this monoallelic nature of mTORC1 mutations, pS6 and p-4EBP1 IHC staining in LOT was significantly lower (H-score) (Supplementary Fig. 2) than what we have previously reported for eosinophilic and vacuolated tumor (EVT), a renal tumor associated with bi-allelic mTORC1 inactivation.³⁰

In the E-chRCC cohort, similar to the findings reported in The Cancer Genome Atlas (TCGA) study,²³ we observed a low frequency of somatic mutations, and TP53 was the only gene with recurrent mutations identified among the 11 cases analyzed (Fig 4A). Copy number analysis in 10 cases identified the multiple chromosomal losses characteristic of chRCC in a subset of 6 cases (60%), including two cases with concurrent WGD or imbalanced chromosome duplication (ICD), a molecular feature previously described in some chRCC.³¹ The remaining 4 cases (40%) had limited or no arm-level chromosomal changes.

The molecular characterization of 3 difficult cases of oncocytic neoplasm NOS (Supplementary Fig. 3) revealed NRAS pathogenic mutation in one case and varied non-specific copy number changes (Fig. 4A and 4C), consistent with their histologic grouping within the current study.

Collecting duct principal cell marker L1CAM distinguishes LOT from E-chRCC

In the human kidney, L1CAM (L1 Cell Adhesion Molecule, also known as CD171) is expressed in the ureteric bud and its derivatives, including the developing collecting ducts and the most distal part of primitive nephrons; in the adult kidney, L1CAM is expressed on the basolateral membrane of principal cells in the collecting ducts and connecting tubules, but not on intercalated cells.^{32, 33} As chRCC and renal oncocytoma are believed to originate from intercalated cells^{34, 35} and LOT shows distinctive IHC features from either entity, we explored the utility of L1CAM as a diagnostic marker to distinguish LOT from E-chRCC.

Consistent with the previous literature, L1CAM staining was seen at the basolateral aspect of most lining cells of medullary collecting ducts, with intermixed scattered cells lacking the staining (discernable at the basal aspect), presumably representing intercalated cells (Fig. 5A). In the cortex, L1CAM was expressed in a subset of distal tubules (Fig. 5B). Remarkably, all 32 LOT cases (100%) that we stained for L1CAM showed diffuse and strong membranous staining, including the only case with focal CD117 staining (Fig. 5C–F and Supplementary Fig. 1). In contrast, 32 of 37 (86%) of stained E-chRCC were negative or only contained scattered rare cells (Fig. 5G); 4 (11%) cases showed focal/patchy staining (Fig. 5H), and only one case (3%) exhibited an interesting heterogeneous staining pattern with juxtaposed, strongly positive areas and negative areas (Fig. 5I). Diffuse and strong membranous staining for L1CAM IHC achieved 100% sensitivity and 100% specificity in distinguishing LOT from E-chRCC in our study cohort. Taken together with previous evidence of distinctive L1CAM expression patterns between principal and intercalated cells, the diffuse and strong L1CAM staining in all LOT tested suggests that this tumor most likely arises from the principal cells of collecting ducts or connecting tubules, a distinct cell of origin from E-chRCC.

Figure 5.

L1CAM is a highly sensitive and specific marker to distinguish LOT from E-chRCC. A) L1CAM immunostain labels the basolateral aspect of the principal cells (PC) (arrow) of medullary collecting ducts; scattered cells lacking the staining are intercalated cells (IC) (arrowhead). B) L1CAM was expressed in a subset of distal tubules in the cortex. C-F) L1CAM stain in all LOT cases showed diffuse and strong membranous staining. G) In E-chRCC, L1CAM stain is mostly negative or only shows rare, scattered cells. H) A small subset of cases shows focal/patchy staining. I) One case showed juxtaposed, strongly positive areas and negative areas.

As GATA3 immunoreactivity has previously been reported in LOT,^{19, 20} we assessed its utility in our cohorts of LOT and E-chRCC. Of the 32 LOT cases tested, 16 (50%) showed diffuse, moderate to strong immunoreactivity, while staining was patchy in 8 (25%) and focal/weak in the remaining 8 (25%) cases (Supplementary Fig. 3). In E-chRCC, 15 (65%) of 23 stained cases were negative for GATA3, however, 8 (35%) tumors showed positive staining, predominantly focal in 6 cases, but patchy and moderate in 2 (9%). Thus, although GATA3 can aid in distinguishing between LOT and E-chRCC, its sensitivity and specificity were lower compared to L1CAM, and some cases exhibited equivocal staining patterns (focal/patchy and weak).

To better understand the potential utility of L1CAM for a broader spectrum of low grade eosinophilic/oncocytic neoplasms, we also explored its expression in a small number of renal oncocytoma (n=9) and oncocytic neoplasm NOS (n=23). All 9 oncocytomas and 22 of 23 oncocytic NOS were negative for staining. Interestingly, one oncocytic NOS case (Onc_NOS2) showed moderate L1CAM positivity with less membranous staining than what is seen in LOT; this case had diffuse CK7 positivity and heterogenous staining for CD117 (Supplementary Fig. 4).

Ultrastructure of neoplastic cells in LOT

All five LOT cases consistently showed the following features under EM: abundant normal-appearing, round to oblong-shaped mitochondria, sparse non-mitochondrial organelles, and scattered intracytoplasmic lumina lined by attenuated microvilli (Fig. 6).

Figure 6.

Ultrastructure of LOT. A) Neoplastic cells have normal-appearing mitochondria and intracytoplasmic lumina with attenuated microvilli (arrow). B) Higher power view of the intracytoplasmic lumina with microvilli.

DISCUSSION

Among low-grade eosinophilic/oncocytic renal neoplasms considered in the differential diagnosis of E-chRCC, the recently described emerging entity LOT shows some histologic and immunohistochemical features that overlap with those of E-chRCC, such as the peri-nuclear clearing and subtle nuclear contour irregularity, as well as the diffuse CK7 positivity that can be seen in a subset of E-chRCC, contributing to the uncertainties regarding its classification among the spectrum of oncocytic neoplasms. We analyzed and compared the morphologic, IHC, and molecular features of well-annotated large cohorts of LOT and E-chRCC to gain a better understanding of their diagnostic and defining features. Furthermore, we identified a novel utility of L1CAM, a marker for principal cells of collecting ducts, as a highly sensitive and specific IHC diagnostic marker to distinguish LOT from E-chRCC, supporting a distinct cell of origin for LOT than E-chRCC and oncocytoma.

Clinically, we found that LOT shows a female predilection, in contrast to the male predominance seen with E-chRCC. While tumors in both cohorts were indolent, there was no disease recurrence after nephrectomy among all LOT patients, whereas two E-chRCC patients developed recurrence or metastasis. Notably, multifocal tumors were found in a small percentage of patients in both groups.

Morphologically, LOT exhibited a homogenous architectural pattern, with all cases predominantly featuring tightly packed small nests with focal edematous or tubular areas. In contrast, for E-chRCC, only 14% of cases showed predominant small-nested architecture, with the large majority displaying classic chRCC or tubular growth patterns. While the frequency of perinuclear halos was similar in LOT and E-chRCC, irregular nuclear contours were vastly more prevalent in E-chRCC. By IHC, we detected limited variation among the LOT cases for CK7 and CD117, with only two tumors not showing the typical staining pattern (i.e. CK7 diffusely positive and CD117 negative). On the other hand, the E-chRCC cohort displayed more variation, with CK7 staining ranging from diffusely positive to negative and CD117 mostly positive, but not infrequently weak or focal. The latter findings make a subset of E-chRCC challenging to separate from LOT when using these two common IHC markers.

L1CAM exhibited remarkably different staining patterns between LOT and E-chRCC, providing a novel ancillary tool to distinguish LOT from E-chRCC in cases with unusual histologic or IHC features. L1CAM is a multidomain membrane glycoprotein of the immunoglobulin superfamily. Initially discovered as a neuronal cell adhesion molecule, L1CAM plays important roles in the development of nervous system, but it can also be upregulated in a variety of malignant tumors and associated with poor outcomes, potentially by promoting tumor cell motility and dissemination.^{36, 37} In the human kidney, prior literature shows that L1CAM is expressed in the ureteric bud and its subsequent derivatives, the collecting ducts and distal portion of connecting tubules. Interestingly, among the two distinct cell types in collecting duct segments, L1CAM is expressed in principal cells but not intercalated cells.^{32, 33} The recent utilization of single-cell technology in kidney reveals a similar but more nuanced picture of cell types within collecting ducts and confirms L1CAM expression in the principal cell cluster.^{38, 39} As the cell of origin for chRCC has been attributed to intercalated cells,^{34, 40} the diffuse and strong membranous staining of L1CAM we observed in all LOT tumors but not in E-chRCC supports the notion that LOT possesses a different cell of origin than E-chRCC, most likely arising from the principal cells of the collecting ducts. Consistent with this difference in cell origin, our limited analysis suggests that L1CAM IHC is also useful in distinguishing LOT from renal oncocytoma.

Molecularly, we observed mTORC1 pathway mutations in 87% of cases in our LOT cohort using a targeted NGS panel, confirming the previous finding that mTORC1 mutations are the primary driver of these tumors. In comparison to other recently described eosinophilic renal neoplasms characterized by mTORC1 pathway mutations, such as eosinophilic solid and cystic carcinoma (ESC) and EVT, which mostly harbor biallelic alterations of TSC1/TSC2 or MTOR genes,^{30, 41} the somatic mutations detected in LOT are most frequently monoallelic gain-of-function mutations of MTOR or RHEB genes. The RHEB gene product is a small GTPase that activates mTORC1 when in its GTP-bound form, and it can be converted to the inactive GDP-bound form by the TSC1/TSC2 complex.⁴² The most common RHEB Y35N mutation identified in this cohort represents a constitutively active mutant,^{43, 44} and evidence from previous studies also supports stimulation of mTORC1 pathway associated with other RHEB mutations (P37L, G63A, and E40V) detected.^45–48 Among all mTORC1 mutations identified in this LOT cohort, MTOR A108F and TSC1 L161Q have not been previously characterized, and their functional impact requires further investigation. On the other hand, the presence of 3 LOT cases lacking mTORC1 mutations suggests there could be alternative mechanisms of tumorigenesis beyond our detection methods.

Additionally, we found that chromosomes 19q and 7 gains are the most frequent copy number alterations in LOT; chromosome 7 gains tend to co-occur with RHEB mutation, while 19q gain often represents a sub-clonal event. Overall, 41% of LOT cases had a flat genome, and no specific arm-level copy number changes appear to be a defining molecular feature of LOT. The one case (T21) in our LOT cohort with biallelic TSC1 alterations and WGD showed nested to sheet-like solid growth and cytoplasmic vacuolization, which focally suggested the morphologic features of ESC or EVT, but it displayed the typical IHC labeling of LOT.

E-chRCC exhibited a molecular profile distinct from LOT. While the number of cases analyzed was limited, we did not detect any mTORC1 mutations, but rather the characteristic chromosomal losses of chRCC in 60% of cases. In the TCGA chRCC cohort,⁴⁹ if one excludes 4 cases consistent with LOT (TCGA-KN-8437, TCGA-KM-8439, TCGA-KM-8441, TCGA-KM-8639),^{10, 16} there is only one case (TCGA-KL-8326) which demonstrated a pathogenic mutation in any of the 4 recurrently mutated mTORC1 pathway genes (MTOR, RHEB, TSC1, TSC2) identified in our LOT cohort – a TSC1 mutation (visualized at cBioPortal http://cbioportal.org^{50, 51}). Thus, while mTORC1 mutations can be seen in various types of renal cell neoplasms, they are highly prevalent in LOT as the primary driver but very uncommon in chRCC. Our molecular analysis of LOT and E-chRCC does highlight a gray zone between E-chRCC with limited or no copy number changes (4/10, 40% of our profiled cohort) and other oncocytic NOS tumors lacking specific molecular findings.

As mTORC1 mutations have also been identified as primary molecular alterations in other newly recognized or emerging renal tumor types (e.g. ESC, EVT, and renal cell carcinoma with fibromyomatous stroma),⁵² we hypothesize that difference in the cell of origin may be one of the mechanisms underlying the broad morphologic spectrum seen with these tumors and renal cell neoplasms occurring in TSC patients. The uniformly strong staining of a principal cell marker L1CAM in all LOT cases but not in E-chRCC or oncocytoma, and the high frequency of mTORC1 mutations in the LOT cohort support that LOT is driven by this pathway and specifically developed from the principal cells of collecting ducts. Previously described ancillary IHC tools for diagnosing LOT, such as positive stain for GATA3,^{19, 20} or an absence of FOXI1 (a transcription factor expressed in intercalated cells),^{8, 10, 15} would also be consistent with a developmental origin from the ureteric bud and a non-intercalated cell (i.e. principal cell) origin we propose for LOT. However, L1CAM expression by IHC shows higher sensitivity and specificity than GATA3 for distinguishing LOT from E-chRCC.

Our EM analysis showed normal-appearing mitochondria in LOT, and tumor cells lacked classic ultrastructural features described in the literature for chRCC (including eosinophilic variant): abundant perinuclear microvesicles and dysmorphic mitochondria (tubulovesicular cristae with cistern-like changes and budding of microvesicle-like structures from the mitochondrial outer membrane).^53–55 Moreover, all 5 LOT tumors exhibited an unusual ultrastructural feature – the presence of intracytoplasmic lumina with attenuated microvilli – which has not been described in most ultrastructural studies of chRCC or renal oncocytoma.^{53, 55} One report of bilateral multifocal renal oncocytomas did identify intracytoplasmic lumina similar to our finding, but histomorphology of some tumors in this case raises suspicion for LOT.⁵⁶ Overall, we propose that the ultrastructural findings in LOT may be unique, and share some similarities with normal principal cells, including sparsity of organelles and attenuated apical/luminal microvilli.⁵⁷

Intriguingly, another recently described indolent renal neoplasm, papillary renal neoplasm of reverse polarity (PRNRP), has been reported to consistently show diffuse membranous staining for L1CAM and GATA3.^58–61 Whether these similar IHC features suggest a shared cell of origin between PRNRP and LOT requires further investigation.

As L1CAM can also be upregulated in various cancer types, the utility of L1CAM in the more general differential diagnosis of renal cell neoplasms requires further investigation. For example, strong L1CAM staining has been reported in 4% of ccRCC, with the same monoclonal antibody (14.10) used herein.⁶² Its expression in other TSC or mTORC1-related tumors such as ESC and EVT also needs clarification. Interestingly, while none of the E-chRCC showed the same diffuse and strong membranous staining as in LOT, L1CAM did highlight scattered cells or focal areas in a portion of cases, suggesting the possibility of either retained principal cells or tumor cells with intermediate features between principal and intercalated cells. One such case showed juxtaposed areas of positive and negative staining. It is worth noting that a checkerboard-like, mutually exclusive L1CAM and FOXI1 or LINC01187 expression pattern has recently been described in hybrid oncocytic tumors associated with Birt-Hogg-Dubé syndrome, suggesting a dual cell origin in this specific clinical scenario.^{63, 64} It remains to be explored whether there is also cell origin heterogeneity in sporadic chRCC cases.

Given the lack of recurrence or metastasis of LOT in all the studies including ours, we propose renaming LOT as “Oncocytic Principal Cell Adenoma of the Kidney”, to better correlate with its essentially benign clinical behavior and reflect its cell of origin. The term “low grade oncocytic tumor” is descriptive in nature and often leads to confusion with general categorical terms such as “low grade oncocytic neoplasm”. The estimated incidence of LOT (0.6%) was just slightly lower than E-chRCC (0.85%) in our consecutive resection cohorts, indicating that it is an important differential consideration within the spectrum of low grade oncocytic renal neoplasms. We suspect that the predominantly monoallelic nature of the mTORC1 mutations in these tumors may underlie its higher incidence than what we observed for EVT or ESC in our practice. Demonstrating consistent morphologic features, LOT can be reliably distinguished from E-chRCC and other low grade oncocytic renal neoplasm based on histomorphology with the aid of a small panel of IHC markers. Although highly prevalent in LOT, mTORC1 somatic mutations are neither specific nor required for the diagnosis of LOT.

In conclusion, our study provides further evidence that LOT is a benign renal neoplasm distinct from E-chRCC. It is characterized by discrete morphologic features, predominantly monoallelic mTORC1 pathway mutations, consistent L1CAM expression, and most likely arises from principal cells of the collecting ducts. We believe that the identification of its unique molecular features and cell of origin, coupled with its indolent clinical behavior, will help promote an improved classification and nomenclature for these tumors.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.