Predicting Oncologic Outcomes in Small Renal Tumors

Abstract

Background:

Most patients diagnosed with renal cancer today present with small renal masses (SRM). Though these patients have a low risk of dying from their disease and many are followed with active surveillance protocols, a small subset of renal cell carcinomas (RCCs) behave aggressively. Knowledge regarding features of aggressive behavior would enable better adoption of active surveillance amongst these patients.

Objective:

We sought to improve prognostic models to predict metastasis-free survival after nephrectomy through focused analyses of clinicopathologic characteristics of SRMs associated with adverse outcomes.

Design, setting, and participants

We identified consecutive patients with surgically resected SRMs (≤ 4 cm) at the University of Texas Southwestern Kidney Cancer Program between 1998 and 2020. In addition, we evaluated the ability of RMs to form tumors when implanted in mice, an indicator of tumor aggressiveness.

Outcome measurements and statistical analysis

We examined the clinicopathologic factors associated with metastasis including prospectively performed BAP1 immunohistochemistry at our Clinical Laboratory Improvement Amendments laboratory. Multivariable Cox proportional hazard regression was used to predict metastasis-free survival.

Results and limitations

A total of 3900 evaluable nephrectomies (from 3674 ethnically diverse patients) were identified, of which 1984 (51%) were SRMs including 1720 RCC. Of these patients with RCC (SRM^RCC), 1576 did not have synchronous or metachronous larger RCCs and among these, 37 (2%) developed metastases. SRM^RCC that metastasized were significantly enriched for aggressive morphologic phenotypes and engrafted in mice at comparable rates as larger metastatic tumors. BAP1 loss remained significantly associated with metastasis-free survival after accounting for TNM (tumor-node-metastasis) stage and SSIGN (stage, size, grade, and necrosis) score in multivariable analysis.

Conclusions:

We identified clinicopathologic features that influence metastasis-free survival for patients with SRM^RCC. If validated independently, these data should assist with patient prognosis and help with active surveillance strategies.

Patient summary:

We report the identification of features of aggressiveness in small renal tumors that influence likelihood of metastases after surgery.

Introduction

Cancer of the kidney and renal pelvis was diagnosed in 76,080 Americans in 2021¹. Most patients with renal cancer today present with stage I tumors (>50%). Though conventional therapy for small renal masses (SRMs [≤ 4 cm in size]) has been nephron-sparing nephrectomy, active surveillance (AS) and less invasive ablative procedures are increasingly being used. The National Comprehensive Cancer Network (NCCN) guidelines recommend AS especially for <2 cm masses and in elderly patients.

Patients with SRMs have been shown to have low risk of dying from their disease². However, a small subset of SRMs metastasize². Current prognostic models^3,4,5,6,7,8 are based on data from all stages and large studies focusing on SRMs are lacking.

Molecular genetics discoveries in renal cancer provide an opportunity to further stratify RCC^9–11. In particular, the BRCA1-associated gene 1 (BAP1), which is mutated in ~15% of clear cell RCC (ccRCC)¹², has consistently been shown to be associated with aggressive features and poor outcome^12,13,10. Using an immunohistochemistry (IHC) assay for BAP1 protein, we confirmed these findings in a large cohort of ~1400 ccRCC patients (hazard ratio [HR] 3.06, 95% confidence interval [CI] 2.28–4.10; p = 6.77 x10⁻¹⁴)^13,14,15, where BAP1 was an independent marker of prognosis in patients with low risk RCC (stage, size, grade, and necrosis [SSIGN] ≤3) Given these findings, BAP1 IHC has been routinely incorporated in our Clinical Laboratory Improvement Amendments laboratory since 2012.

Information about prognostic variables specifically for RCC that present as SRM (SRM^RCC) is limited. The Renal Task Force of the National Cancer Institute convened a workshop (January 23-24, 2020, Bethesda, MD, USA) with the goal of advancing the state-of-the-science and develop prognostic biomarkers to individualize treatment for patients with SRMs. Herein, we evaluate the incidence and clinicopathologic features of SRM at the University of Texas Southwestern (UTSW) Kidney Cancer Program to identify the determinants of aggressive biology.

Patients and methods

Patient selection

The study was conducted with approval by the UTSW Institutional Review Board (STU 022015-015). We searched our institutional RCC database (Kidney Cancer Explorer [KCE]), and identified a total of 3968 consecutive partial and/or radical nephrectomies from 3729 patients (between 1998-2020 at our institutional university hospital and 2013-2017 at the affiliated county Parkland Hospital, Dallas) with a diagnosis of renal neoplasm.

Sixty-eight RCCs including eight SRM ^RCC cases with metastasis were excluded from further analyses due to a history of prior RCCs resected at outside institutions with either limited clinical/pathologic data or larger RCCs.

Data collection

Associated clinicopathologic and follow-up information was updated as of May 11, 2022. Patients with nephrectomies performed at outside institutions with histology slides reviewed at the UTSW who subsequently received therapy at our institution were included (n = 97 including 21 SRMs). Tumors ≤ 4 cm in size were considered SRMs. BAP1 IHC status¹³ for all ccRCC cases was extracted from the pathology reports. For this study, tumors were subtyped according to the World Health Organization (WHO) histologic classification and grading systems in vigor at the time. The tumors were restaged based on the 2018 American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) classification for pathologic staging. Primary outcomes assessed were metastasis (presence of a distant lymph node and/or organ metastasis) and death.

In patients with multiple renal tumors, tumor information corresponding to the earliest tumor was extracted. When multiple tumors (removed within 6 months of each other) were present at first diagnosis, the characteristics of the largest RCC (dominant tumor) were recorded. Patients were considered to have SRM^RCC if they had no synchronous or metachronous RCC > 4 cm; recurrent tumors were excluded. Rare instances of patients with more than one renal tumor, including a tumor >4 cm, were included if they developed metastases which could be traced back based on histological analyses to the SRM.

Data in KCE were supplemented through comprehensive review of clinical data and pathology slides. Electronic medical records of patients with SRM^RCC were comprehensively evaluated for diagnosis and treatment of prior RCCs that may have occurred at outside institutions.

Hematoxylin and eosin-stained tumor sections of available SRMs with metastasis were reviewed. Morphologic ontology categorization was captured by two genitourinary pathologists (P.K. and Q.C.) as described previously¹⁶.

Dating back to 2008, our Kidney Cancer Program has generated a comprehensive resource through the orthotopic implantation of tumors into immunocompromised mice^17,18. Data of the implanted and engraftment tumors were collected from KCE and other resources.

Statistical methods

All features in the outcome data retrieved from KCE were summarized with event frequencies, percentages, or median and interquartile ranges (IQR), as appropriate. When not reported, sarcomatoid features were considered absent in tumors of grade 1-3. BAP1 expression (ccRCC only), was dichotomized as negative/loss versus positive/retained. The date of first diagnosis was the date of nephrectomy or prior biopsy of the renal mass. Documentation of metastasis was based on date of biopsy or, if not available, date of initiation of therapy (systemic therapy, surgery, or radiation) or unequivocal imaging confirming diagnosis of distant metastasis. Metastasis-free survival (MFS) was calculated from the date of first diagnosis to the date of documented first metastasis, last follow-up or death. Overall survival (OS) was calculated from the date of first diagnosis to the date of last follow-up or death from any cause. Survival of patients who were lost to follow-up was censored at the date of last contact. MFS and OS were determined by Kaplan-Meier estimates and visualized using the R package survival (version 3.2-13). Hazard ratios (HR), 95% confidence intervals (CI) and p-values were obtained from Cox regression analyses by the R packages rms (version 6.2-0). The Cox proportional hazard model (including multivariables) was utilized to evaluate the association between time to metastasis events and each clinicopathologic feature by the R packages rms. Unless indicated, p-values are two-sided without adjusting for multiple comparisons. Statistical analyses and visualization were conducted using R (version 4.2.0).

Results

Spectrum of histologic subtypes in SRMs

A total of 3900 consecutive evaluable partial and/or radical nephrectomies from 3674 patients were identified. Approximately half of the nephrectomies were ≤4 cm SRMs (SRMs/Total; 1984/3900 = 51%). Among the 1984 SRMs in 1887 patients, 1721 (87%) were malignant (n=1 Wilms tumor and 1720 RCCs), 130 (6.6%) were renal neoplasms of low/undetermined malignant potential (UMP), and 133 (6.7%) were benign tumors (Table 1). Amongst the 1720 small RCCs from 1655 patients, 1576 had SRM^RCC as the dominant RCC with no > 4 cm synchronous or metachronous RCC. For this study we focused our evaluation on the SRM^RCC cohort.

Table 1.

Distribution of SRM histologic subtypes from 1887 patients

Histology	No. of cases	% of category	% of total
ccRCC	1284	75	65
PRCC	270	16	14
ChRCC	82	4.8	4.1
Unclassified RCC	59	3.4	3.0
Acquired cystic disease-associated RCC	12	0.70	0.61
TRCC	8	0.47	0.40
Mucinous tubular and spindle cell carcinoma	1	0.06	0.05
Tubulocystic RCC	4	0.23	0.20
Nephroblastoma (Wilms tumor)	1	0.06	0.05
Total Malignant	1721		87
ccPRCC	81	62	4.1
Multilocular cystic clear cell renal cell neoplasm of low malignant potential	27	21	1.4
Renal Oncocytic Neoplasm	22	17	1.1
Total UMP	130		6.6
Oncocytoma	82	62	4.1
Angiomyolipoma	40	30	2.0
Cystic Nephroma	7	5.3	0.35
Metanephric adenoma	4	3.0	0.20
Total Benign	133		6.7

RCC: renal cell carcinoma; ccRCC: clear cell RCC; PRCC: Papillary RCC; ChRCC: Chromophobe RCC; TRCC: MiT family translocation RCC; ccPRCC: clear cell papillary RCC; Undetermined/low: Undetermined malignant potential

Baseline Clinical Characteristics of Study Cohort

We first compared the clinicopathologic features of SRM^RCC and RCC larger than 4 cm (1576 and 1768 patients respectively) (Supplementary Table 1). When more than one malignant tumor was diagnosed in a patient, the first RCC was used as a reference (see Methods for more details). Patient demographics as well as tumor grade and stage distributions are representative of most large referral centers^14,7. Selected clinicopathological features of the 1576 patients with SRM^RCC are shown in Supplementary Table 1.

Patients with SRM^RCC had a slightly younger median age at nephrectomy with less male predilection when compared to patients with larger RCCs. The frequency of papillary RCC (PRCC) was higher in the SRM^RCC subgroup. Compared to RCC > 4 cm, SRM^RCC was infrequently associated with adverse risk variables such as high nuclear grade (G3-4), sarcomatoid and rhabdoid features, tumor necrosis, presence of lymphovascular invasion (LVI), pT3-4, pN1, synchronous metastasis (M1), and TNM stage III-IV (all with Chi-squared p<0.001) (Supplementary Table 1).

Small RCCs infrequently metastasize

The median duration of follow-up for the entire cohort of 2808 RCC patients without metastasis was 2.7 years (IQR 0.6-5.9 years) and 3.0 years (IQR 0.6-6.1 years) for 1539 SRM^RCC patients without metastasis. Among the 1576 patients with SRM^RCC, 37 patients developed metastatic disease including 11 patients with metastases at the time of nephrectomy (M1). This contrasts with 499 of 1768 patients with larger tumors, including 278 having M1 disease at diagnosis. Overall, the frequency of metastasis was significantly lower in SRM^RCC than in patients with large RCCs. Metastatic SRM^RCC (SRM^mRCC) constituted 6.9% of the metastatic RCCs though SRM^RCC comprised 47% of the cohort.

As expected, SRM^RCC, had significantly longer metastasis-free survival (MFS) and overall survival (OS) as compared to large RCC (MFS: HR, 0.07; 95% CI 0.05-0.10; p<0.001; OS: HR, 0.40; 95% CI 0.33-0.47; p<0.001) (Figure 1).

Figure 1.

Kaplan-Meier survival curves showing the association between size and (A) metastasis-free survival (HR 0.07, 95%CI 0.05-0.10, log-rank p < 0.001); (B) overall survival (HR 0.40, 95%CI 0.33-0.47, log-rank p < 0.001) in the patients diagnosed with RCC at UTSW (n=3344). CI = confidence interval; HR = hazard ratio; RCC = renal cell carcinoma.

SRM engraft at significantly lower rates than large RCCs

Patient derived tumorgraft models, which involve the implantation of patient tumor samples into immunocompromised mice, recapitulate the aggressiveness of RCC. We have previously shown that stable engraftment of primary tumors predicts for metastatic potential and poor patient survival^17,18. We asked if there was a difference in engraftment rates based on the size of the primary tumor. During 2008-2019, RCC samples from 708 nephrectomies were implanted in mice, including 113 from SRMs. Unsurprisingly, we saw lower engraftment rates from SRM^RCC (3.5%) compared to large RCCs (18%) p<0.001 (Figure 2). Interestingly, the rate of engraftment was similar for tumors from patients who developed metastasis irrespective of the size of the primary tumor (75% for SRM^mRCC vs. 71% for metastatic large RCCs).

Figure 2.

Metastatic SRMs engraft at rates comparable to larger metastatic tumors despite lower overall engraftment rates of non-metastatic SRMs. SRM = small renal mass

Clinico-pathologic features of metastasizing SRM^RCC

We then evaluated the clinicopathologic characteristics of SRM^RCC with and without metastasis (Table 2). We asked if there was a positive association between metastasis and an increase in size in SRM^RCC. No metastasis was observed in RCCs less than 2 cm, with the exception of an acquired cystic disease-associated RCC (Supplemental Figure 1). The median size in SRM^mRCC was 3 cm [IQR 2.7-3.6]. Patients with SRM^mRCC tended to be older men (66 years [IQR 60-70] vs. 59 years [IQR 49-67]) and tumors more frequently exhibited adverse histopathologic features such as high grade (G3-4), sarcomatoid or rhabdoid features, tumor necrosis, LVI, pT3, pN1, and TNM stage III-IV (Table 2). Amongst patients with clear cell SRM^RCC (SRM^ccRCC), the most common subtype, loss of BAP1 occurred more frequently in patients with metastasis (28% vs 6.9%; Fisher’s exact p<0.001).

Table 2.

Patient and clinical characteristics of SRM^RCC stratified by whether they developed metastases (n = 11 metastatic at diagnosis; n = 26 metastatic after diagnosis)

N = 1576		Metastasis Events/Total* (%)	Hazard Ratio (95% CI)	Cox p
Age at Nephrectomy	per 5 years		1.30 (1.12, 1.51)	<0.001
Gender	Female	9/674 (1.3)	Reference	0.025
	Male	28/902 (3.1)	2.36 (1.11, 5.00)
Race	White	29/1273 (2.3)	Reference	0.7
	Black	5/198 (2.5)	1.02 (0.40, 2.64)
	Other	2/45 (4.4)	1.85 (0.44, 7.76)
Ethnicity	Non Hispanic	34/1295 (2.6)	Reference	0.098
	Hispanic	2/229 (0.87)	0.30 (0.07, 1.25)
Focality	Unifocal	29/1403 (2.1)	Reference	0.3
	Multifocal	5/102 (4.9)	2.03 (0.79, 5.25)
Surgical approach	Radical nephrectomy	17/278 (6.1)	Reference	<0.001
	Partial nephrectomy	20/1260 (1.6)	0.26 (0.14, 0.49)
Size	per 1 cm		2.27 (1.45, 3.54)	<0.001
Histology	ccRCC	29/1198 (2.4)	Reference	0.6
	PRCC	4/230 (1.7)
	ChRCC	0/76 (0.00)	0.82 (0.38, 1.80)
	Unclassified RCC	1/54 (1.9)
	Others	3/18 (17)
Nuclear Grade	1/2	10/1068 (0.94)	Reference	<0.001
	3	20/393 (5.1)	6.16 (2.88, 13.2)
	4	6/24 (25)	44.3 (15.8, 124)
Sarcomatoid features	Not identified	35/1569 (2.2)	Reference	<0.001
	Present	2/6 (33)	35.6 (8.35, 152)
Rhabdoid features	Not identified	32/1500 (2.1)	Reference	<0.001
	Present	3/8 (38)	30.5 (9.21, 101)
Tumor Necrosis	Not identified	21/1069 (2.0)	Reference	<0.001
	Present	14/155 (9.0)	4.35 (2.21, 8.56)
LVI	Not identified	24/1348 (1.8)	Reference	<0.001
	Indeterminate	0/23 (0.00)	6.30 (2.82, 14.1)
	Present	8/60 (13)
pT	1	24/1464 (1.6)	Reference	<0.001
	3	13/112 (12)	8.47 (4.29, 16.7)
pN	0/X	35/1572 (2.2)	Reference	<0.001
	1	2/4 (50)	40.6 (9.69, 170)
TNM Stage	1	17/1456 (1.2)	Reference	<0.001
	3	9/109 (8.3)	8.77 (3.88, 19.8)
	4	11/11 (100)	933 (313, 2779)
BAP1 Status (ccRCC)	ccRCC_BAP1+	21/660 (3.2)	Reference	<0.001
	ccRCC_BAP1-	8/55 (15)	4.69 (2.08, 10.6)
	Non-ccRCC	8/378 (2.1)	0.61 (0.27, 1.38)

^*Total SRM^RCC excluding missing values for each feature.

Clinico-pathologic features fail to adequately capture aggressive SRM^RCC compared to larger metastasizing tumors

We next evaluated clinicopathologic variables in our cohort of 536 RCCs with metastases stratified based on SRM status (Supplemental Table 2). Patients with SRM^mRCC were older. Aggressive clinicopathologic features such as high grade (G3-4), sarcomatoid change, tumor necrosis, LVI, pT3-4, pN1, and TNM stage III-IV (Supplemental Table 2) were not as frequently present when compared to large RCC with metastasis. However, there were no significant MFS or OS differences amongst patients with metastasis based on tumor size (Figure 3). Four (4/37) SRM^mRCC did not have any aggressive clinical features, in contrast to only 9 (9/499) large RCCs with metastasis. These data suggest that current prognostic variables may not capture biologically aggressive SRM^RCC to the same degree as in large metastatic RCCs.

Figure 3.

Kaplan-Meier survival curves showing the association between size and (A) metastasis-free survival (HR 0.76, 95%CI 0.54-1.06, log-rank p 0.1); (B) overall survival (HR 1.00, 95%CI 0.65-1.53, log-rank p 1) in the patients diagnosed with metastatic RCC at UTSW (n=536).

SRM^ccRCC that metastasize exhibit aggressive architectures

We asked if SRM^mRCC has unique morphologic features. We have previously described the morphologic phenotypes of aggressive behavior^16,19,20. We reviewed metastatic SRM^ccRCC with available tissue (n=26) and comprehensively characterized their morphologies. We compared their architectural composition to the architecture composition of the previously reported non-metastatic SRM^ccRCC cohort (n=233)¹⁶. We observed that the presence of aggressive morphologic phenotypes such as alveolar, thick trabecular/insular, papillary/pseudopapillary and solid sheet architectures in SRM^ccRCC significantly increased the likelihood of metastasis (Table 3). Two of the four SRM^mRCC that all established aggressive clinicopathologic features were ccRCC. One showed aggressive architectural patterns and the other was an outside case not available for review. Furthermore, we found that metastatic SRM^ccRCC had higher intratumoral heterogeneity than non-metastatic SRM^ccRCC (mean difference of the number of architectural patterns is 0.85 [95%CI; 0.29-1.41; t-test p=0.004]). These data suggest that aggressive SRMs are morphologically and biologically akin to larger RCCs with metastases.

Table 3.

Architectural patterns present in SRM^ccRCC stratified by whether the patient ever developed metastasis

N = 259	Pattern present	Metastasis Events/Total (%)	Hazard Ratio (95% CI)	Cox p
Microcystic	N	24/155 (15)	Reference	0.003
	Y	2/104 (1.9)	0.11 (0.03, 0.47)
Tubular/Acinar	N	24/181 (13)	Reference	0.018
	Y	2/78 (2.6)	0.18 (0.04, 0.75)
Bleeding follicles	N	23/176 (13)	Reference	0.022
	Y	3/83 (3.6)	0.25 (0.07, 0.82)
Compact small nests	N	11/65 (17)	Reference	0.034
	Y	15/194 (7.7)	0.43 (0.20, 0.94)
Large nests	N	9/132 (6.8)	Reference	0.093
	Y	17/127 (13)	2.00 (0.89, 4.48)
Alveolar	N	11/201 (5.5)	Reference	<0.001
	Y	15/58 (26)	5.16 (2.37, 11.2)
Papillary/Pseudopapillary	N	21/239 (8.8)	Reference	0.043
	Y	5/20 (25)	2.74 (1.03, 7.29)
Thick Trabecular/Insular	N	9/214 (4.2)	Reference	<0.001
	Y	17/45 (38)	11.8 (5.23, 26.5)
Solid sheet	N	7/221 (3.2)	Reference	<0.001
	Y	19/38 (50)	21.9 (9.16, 52.5)

Each tumor could have more than one pattern. Numbers and percentages represent the number of cases that had the morphologic pattern.

Multivariable analysis identifies predictors of disease progression including BAP1

Given that ccRCC is the most common subtype, we asked if inclusion of genomic biomarkers such as BAP1 status along with clinicopathologic features could predict metastasis in SRM^RCC. Multivariable analyses adjusting for TNM stage showed that BAP1 status was independently associated with time to metastasis with a p-value <0.02 (BAP1 loss HR, 3.05; 95% CI, 1.30-7.15) (Table 4). Furthermore, BAP1 status remained significant after adjusting for SSIGN score (Table 5). Taken together, these data provide support that BAP1 has prognostic implications that can complement current prognostic parameters and assist in guiding clinical decision making in patients with SRM^RCC.

Table 4.

Multivariable Cox proportional hazards model for time to metastasis in SRM^RCC, controlling for TNM stage at diagnosis

N=1093*	Metastasis Events / Total (%)	Hazard Ratio (95% CI)	Cox p
BAP1 IHC			0.02
ccRCC BAP1+	21/660 (3.2)	Reference
ccRCC BAP1−	8/55 (15)	3.05 (1.30, 7.15)
Non-ccRCC	8/378 (2.1)	0.85 (0.37, 1.98)

^*SRM^RCC with available BAP1 status and TNM stage.

Table 5.

Hazard ratios for time to metastasis and p-values for BAP1 expression after adjusting for SSIGN score in SRM^RCC

N = 791*	Metastasis Events / Total (%)	Hazard Ratio (95% CI)	Cox p
BAP1 IHC			0.003
ccRCC BAP1+	19/525 (3.6)	Reference
ccRCC BAP1−	8/47 (17)	3.58 (1.53, 8.35)
Non-ccRCC	7/219 (3.2)	0.72 (0.30, 1.74)
SSIGN score		1.94 (1.71, 2.19)	<0.001

^*SRM^RCC with available BAP1 status and SSIGN score.

Discussion

SRMs represent the majority of the renal tumors diagnosed today. However, there are limited data and actionable decision-making tools to guide appropriate management. In the present study, using one of the largest single institutional retrospective experiences, we observed that BAP1 loss remains prognostic for time to metastasis after controlling for TNM stage. Thus BAP1 status complements TNM stage in patient prognosis, and they together may better risk stratify patients with SRM and aid with clinical management. Furthermore, BAP1 status was also independently prognostic of the widely used SSIGN stratification system.

Extending published literature, we found that approximately half of the nephrectomies performed at our institution were SRMs of which 87% were malignant. Prior data suggest that as many as 20% of SRMs are benign²¹. The higher rates of malignancy in our series may be due to a focus on surgical resections. SRMs with benign radiological appearance and/or pathology on a needle biopsy are managed non-surgically at our institution. Over the last several years, a magnetic resonance imaging (MRI)-based, clear cell likelihood scoring system (ccLS)²² that predicts the likelihood of renal mass to be of the more aggressive clear cell subtype is being utilized at our institution and may have further reduced surgeries of benign/indolent SRMs.

We found that most resected SRM^RCC cases are indolent and only 2.3% developed metastasis. No metastases were detected in patients diagnosed with benign or UMP SRM in our cohort. We found a positive association of metastasis with increase in size and negligible risk of metastasis in RCCs < 2 cm. Similar findings have been published in SRM²³ with rate of metastasis reported from 0-6.0%^24–27. These data support the notion that most SRMs especially those < 2 cm in size may be managed by AS approaches, however, a small number of SRMs that are > 2 cm behave aggressively highlighting the need for biomarker development.

There has been a paucity of data for predicting metastasis in SRMs²⁴. We did not have a single chromophobe SRM^RCC with metastasis, a finding that is consistent with its more indolent behavior. Consistently, data from other large institutions show that risk factors for metastasis in chromophobe RCCs include large size, small vessel invasion, and sarcomatoid differentiation^28,29. These findings suggest that in SRM, it may be less critical to distinguish chromophobe RCC from oncocytoma, its benign morphologic mimicker. In contrast, SRMs^RCC with aggressive histologies (Fumarate hydratase-deficient RCC and Translocation RCC) developed metastasis.

The frequency of high nuclear grade (G3-4), necrosis, sarcomatoid, or/and rhabdoid features and high TNM stage (III-IV) were observed in 72% (26/36), 40% (14/35), 5.4% (2/37), 8.6% (3/35), and 54% (20/37) of patients with SRM^mRCC respectively. At least one of these adverse histologic features were present in almost 90% (33/37) of the SRM^RCC with metastasis. These clinical pathologic features can help stratify the risk of metastasis following surgery and guide clinical decision making.

Although SRM^mRCC were more frequently associated with known adverse histopathologic features than non- SRM^mRCC, the frequency was less than that observed in large RCCs with metastases. Inasmuch, as larger tumors are likely to overgrow vascular supply resulting in tumor necrosis, which may promote clonal evolution and increased fitness. Thus, the frequent association with higher grade and necrosis may reflect longer evolution. Our data suggest that metastatic competence and tumor growth may be, at least in some cases, unlinked. Provocatively, OS and MFS was not significantly different for metastatic RCCs when grouped into small and large. Consistently, we found that SRM^RCC that metastasized had the rate of engraftment in mice (a surrogate for tumor aggressiveness) comparable to larger metastatic tumors¹⁸.

Data from our group and others have shown that after VHL, ccRCCs are most frequently mutated for PBRM1, SETD2 and BAP1, and that BAP1 and PBRM1 loss are largely mutually exclusive¹³. Furthermore, unlike PBRM1-deficient tumors exhibit aggressive histopathological features and worse outcomes^10,15. In the current study we tested this further in an independent cohort using a BAP1 assay that was performed prospectively, as samples were collected, in our CLIA lab. The current study shows that BAP1 loss is independently prognostic in SRM^ccRCC and this remains significant after adjusting for SSIGN score. These data are consistent with mouse models for Bap1- and Pbrm1- deficient ccRCCs that support the notion that BAP1 and PBRM1 are not only necessary for ccRCC development but are also determinants of tumor grade^30,31.

The finding that BAP1 is independently prognostic in SRM^ccRCC may have practical application in particular since BAP1 status on biopsy is particularly accurate³². In our study evaluating patients with paired biopsies and subsequent resection of a renal mass, we found that while our ability to diagnose and render a specific histologic subtype on the biopsy was excellent, biopsies underestimated histologic grade, necrosis, and sarcomatoid/rhabdoid features³². Amongst all prognostic variables evaluated, BAP1 status in ccRCC had the highest agreement³². Thus, needle biopsy can accurately represent the entire tumor for some (truncal) prognostic biomarkers such as BAP1, which could be used to better risk stratify patients with SRMs.

We recently established a comprehensive morphologic ontology and described the morphologic phenotypes of aggressive behavior in ccRCC based on cytology, architecture, and the tumor microenvironment¹⁶. Architectures could be broadly categorized into indolent (microcystic, tubular/acinar, bleeding follicle, and compact small nests) and aggressive (alveolar, papillary/pseudopapillary, thick trabecular/insular, and solid sheet)^16,20. We expanded these analyses in our current cohort of SRM^ccRCC and confirmed that the presence of aggressive morphologic phenotypes in SRMs significantly increased the likelihood of metastasis. The morphologic phenotypes of aggressive SRM^ccRCC were comparable to those seen in large aggressive RCC.

Our data suggests that aggressive SRM^RCC have unique clinicopathologic features and biology. Thus, to understand its molecular biology, we have undertaken a multi-institutional comprehensive collaboration.

Our study is not without limitations, which include its retrospective nature and associated non standardized protocol for follow-up. In addition, our study focuses only on cases that underwent surgical resection and may not reflect the entire population of SRM patients, which includes those perceived to have lower risk disease that are followed by AS and never undergo surgery³³. In addition, the metastatic cohort used as a reference includes only those patients that had surgery for their primary tumors. We are also limited by follow-up with a median of 2.7 years. A small percentage of patients did not have tissue available for analyses. Nevertheless, to our knowledge this is one of the first studies in the literature evaluating determinants of metastases in a large cohort of SRMs.

Conclusions

Although these conclusions will require external validation, the approaches proposed may prove valuable in tailoring surveillance and informing treatment selection for patients with SRMs.