Advertisement

Histologic Growth Patterns in Clear Cell Renal Cell Carcinoma Stratify Patients into Survival Risk Groups

Open AccessPublished:January 12, 2022DOI:https://doi.org/10.1016/j.clgc.2022.01.005

      Abstract

      Introduction

      Genomic and morphologic heterogeneity in clear cell renal cell carcinoma (ccRCC) presents a barrier to prognostication and treatment decisions. Data from pathology are used with clinical markers to predict disease progression after nephrectomy. However, determining the risk of cancer recurrence, and survival with metastatic cancer remains challenging. Recently, analysis of histologic growth patterns (HGP) in ccRCC revealed promising associations with survival outcomes.

      Methods

      To investigate whether HGPs can be used to predict overall survival (OS) after nephrectomy, we examined 24 HGPs in primary tumors of 147 patients that included 107 patients with metastatic disease.

      Results

      The median number of HGPs per case was 5 and was greater in metastatic and larger tumors. After adjustment for 6 pathologic and demographic variables, HGPs were significantly associated with OS post nephrectomy. Small nests, expansile nests and nests with high nuclear to cytoplasmic ratio were associated with favorable outcomes; while spindled low grade, fused nests/solid sheets, rhabdoid, and sarcomatoid patterns were associated with unfavorable outcomes. A 3-tiered and a 2-tiered risk model were developed based on combinations of HGPs. The models performed equally well as WHO/ISUP nucleolar plus necrosis grade (necrosis grade), and better than WHO/ISUP nucleolar grade alone in predicting OS at the median OS of 6 years. Pairwise correlations between HGPs revealed 2 tumor evolutionary branches that differed in risk of metastatic disease: one with mesenchymal differentiation, and other with epithelial tubulopapillary differentiation. While 44 of 107 (41%) patients with metastatic ccRCC displayed evidence of mesenchymal differentiation, mesenchymal features were only observed in 1 of 40 (3%) patients without evidence of metastatic disease.

      Conclusion

      These findings suggest that HGPs may provide a novel path to refine the estimation of OS after nephrectomy and to determine the molecular basis of tumor evolution.

      Keywords

      Introduction

      Clear cell renal cell carcinoma (ccRCC) is the most common form of kidney cancer and 12th most common cause of cancer related deaths in the United States.

      American Cancer Society. Cancer Statistics Center. (Accessed at: June 30, 2021 Accessed from: https://cancerstatisticscenter.cancer.org/?_ga=2.180760999.1670122945.1625274963-472095398.1625274963#!/.)

      Up to 30% of the patients present with metastatic disease at diagnosis
      • Siegel RL
      • Miller KD
      • Jemal A.
      Cancer statistics.
      and an additional 30% of patients with localized disease progress to metastasis
      • Dabestani S
      • Thorstenson A
      • Lindblad P
      • Harmenberg U
      • Ljungberg B
      • Lundstam S.
      Renal cell carcinoma recurrences and metastases in primary non-metastatic patients: a population-based study.
      ,
      • Rabinovitch RA
      • Zelefsky MJ
      • Gaynor JJ
      • Fuks Z.
      Patterns of failure following surgical resection of renal cell carcinoma: implications for adjuvant local and systemic therapy.
      with a median disease-free survival (DFS) of approximately 2.2 years.
      • Motzer RJ
      • Tannir NM
      • McDermott DF
      • et al.
      Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma.
      Predicting the risk of metastatic progression after surgery or overall survival (OS) in individual patients is challenging and currently relies solely on pathology variables, such as tumor stage, tumor size, extrarenal extension and renal vein thrombosis; or laboratory parameters as in the International Metastatic Disease Consortium (IMDC) calculator and the UCLA Integrated Staging System (UISS).
      • Zisman A
      • Pantuck AJ
      • Wieder J
      • et al.
      Risk group assessment and clinical outcome algorithm to predict the natural history of patients with surgically resected renal cell carcinoma.
      While pathologic evaluation remains the gold standard for subtyping of RCCs, it does not play a major role in assessing OS in patients with a new diagnosis of ccRCC. Because of considerable morphologic heterogeneity within each patient's tumor, the World Health Organization/International Society of Urologic Pathology (WHO/ISUP) grading scheme is unable to capture the aggressive behavior of the entire tumor. The contemporary clinical WHO/ISUP grading system for ccRCC is based primarily on the size of nucleoli for grades 1-3 and does not incorporate other histopathologic features. The majority of ccRCCs (64%-85%) are assigned to intermediate WHO/ISUP grade categories of 2 and 3
      • Delahunt B.
      Advances and controversies in grading and staging of renal cell carcinoma.
      and 7%-20% of ccRCC are grade 4 tumors.
      • Delahunt B
      • McKenney JK
      • Lohse CM
      • et al.
      A novel grading system for clear cell renal cell carcinoma incorporating tumor necrosis.
      ,
      • Dagher J
      • Delahunt B
      • Rioux-Leclercq N
      • et al.
      Clear cell renal cell carcinoma: validation of world health organization/international society of urological pathology grading.
      Grade 4 is defined by sarcomatoid or rhabdoid differentiation and syncytial tumor giant cells in addition to the nucleolar size and nuclear pleomorphism.
      • Delahunt B
      • Cheville JC
      • Martignoni G
      • et al.
      The international society of urological pathology (ISUP) grading system for renal cell carcinoma and other prognostic parameters.
      ,
      • Moch HHP
      • Ulbright TM
      • Reuter VE
      WHO classification of tumors of the urinary system and male genital organs.
      The intermediate WHO/ISUP grade categories poorly predict disease progression
      • Dagher J
      • Delahunt B
      • Rioux-Leclercq N
      • et al.
      Clear cell renal cell carcinoma: validation of world health organization/international society of urological pathology grading.
      ,
      • Khor LY
      • Dhakal HP
      • Jia X
      • et al.
      Tumor necrosis adds prognostically significant information to grade in clear cell renal cell carcinoma: a study of 842 consecutive cases from a single institution.
      ,
      • Kuthi L
      • Jenei A
      • Hajdu A
      • et al.
      Prognostic factors for renal cell carcinoma subtypes diagnosed according to the 2016 WHO renal tumor classification: a study involving 928 patients.
      and the inclusion of histologic and cytologic features in grade 4 tumors improves the prognostic accuracy, demonstrating that features other than nucleolar grade bear prognostic information.
      The role of histologic growth patterns (HGPs) in predicting prognosis has been demonstrated for some morphologic variants of conventional ccRCC.
      • Kryvenko ON
      • Roquero L
      • Gupta NS
      • Lee MW
      • Epstein JI.
      Low-grade clear cell renal cell carcinoma mimicking hemangioma of the kidney: a series of 4 cases.
      • Williamson SR
      • MacLennan GT
      • Lopez-Beltran A
      • et al.
      Cystic partially regressed clear cell renal cell carcinoma: a potential mimic of multilocular cystic renal cell carcinoma.
      • Montironi R
      • Lopez-Beltran A
      • Cheng L
      • Galosi AB
      • Montorsi F
      • Scarpelli M.
      Clear cell renal cell carcinoma (ccRCC) with hemangioblastoma-like features: a previously unreported pattern of ccRCC with possible clinical significance.
      Furthermore, necrosis has been identified as an adverse prognostic indicator of cancer specific survival when incorporated into the grading system.
      • Delahunt B
      • McKenney JK
      • Lohse CM
      • et al.
      A novel grading system for clear cell renal cell carcinoma incorporating tumor necrosis.
      ,
      • Khor LY
      • Dhakal HP
      • Jia X
      • et al.
      Tumor necrosis adds prognostically significant information to grade in clear cell renal cell carcinoma: a study of 842 consecutive cases from a single institution.
      ,
      • Dagher J
      • Delahunt B
      • Rioux-Leclercq N
      • et al.
      Assessment of tumour-associated necrosis provides prognostic information additional to World Health Organization/International Society of Urological Pathology grading for clear cell renal cell carcinoma.
      ,
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      More recently a comprehensive and systematic analysis of HGPs in ccRCC was reported by 2 separate groups.
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      • Cai Q
      • Christie A
      • Rajaram S
      • et al.
      Ontological analyses reveal clinically-significant clear cell renal cell carcinoma subtypes with convergent evolutionary trajectories into an aggressive type.
      • Kapur P
      • Christie Alana
      • Rajaram Satwik
      • et al.
      What morphology can teach us about renal cell carcinoma clonal evolution.
      In these studies, the authors provided evidence for the prognostic significance of HGPs using the DFS endpoint after nephrectomy. Prognostic significance was also observed with cancer specific survival outcomes. In addition, one of the groups proposed an evolutionary relationship between HGPs that may lead to tumor progression. Both HGP studies demonstrate that analysis of the primary tumor bears information on patient survival when the cancer becomes metastatic.
      With the goal of better understanding the association between HGPs and patient outcomes in ccRCCs, we defined 24 HGPs and assessed whether these HGPs (1) predict OS of patients after nephrectomy and (2) can be assembled into an evolutionary tree of cancer progression.

      Materials and Methods

      Case Selection

      In this retrospective study, consecutive patients seen with the diagnosis of metastatic ccRCC from 1998 to 2018, that had nephrectomy performed in house with pathology slides available for review were selected. In addition, 40 consecutive cases with non–metastatic disease who survived at least 3 years were selected from the pathology archives with no evidence of disease progression/recurrence after nephrectomy. The diagnosis in all cases was confirmed as ccRCC based on a combination of established morphologic and immunohistochemical (IHC) features and/or the presence of VHL gene mutations. Clinical data for all cases was collected. OS was defined by time from nephrectomy to time to death from any cause with censoring at loss to follow-up. The study was performed in compliance with institutional review board (IRB #00067518).

      Analysis of Histologic Growth Patterns

      All available Hematoxylin and Eosin (H&E) and IHC stained slides for cases were systematically reviewed by an anatomic pathologist with expertise in genitourinary pathology (D.S). WHO/ISUP grade (hereafter referred to as nucleolar grade [NG]) and any extent of established architectural and cytologic features that all pathologists are familiar with observed at low power (40x magnification) and necrosis (hereafter referred to as histologic growth patterns (HGPs) for the purpose of this study) were recorded as present or absent. HGPs comprised of architectural patterns, cytologic features, and tumor associated features appreciated in a given cases regardless of reports in prior studies. HGPs were defined per established histopathology criteria. Quantitative assessment of the patterns was not performed. The HGPs were evaluated using an Olympus BX41 light microscope.

      Statistical Methods

      Analysis of HGPs and Clinical Outcomes

      The number of HGPs was compared between metastatic and non–metastatic cohorts using a t test; and the associations between HGPs and metastatic diagnosis were tested using Barnard's Exact Test.
      • Barnarad G.
      A new test for 2 x 2 tables.
      A 3-year progression landmark analysis was conducted to assess the robustness of conclusions comparing metastatic versus non–metastatic cases. The association between each HGP and OS was assessed using a Cox proportional hazards model adjusting for age, sex, race, tumor size, extra-renal extension, renal vein thrombus, necrosis (except when necrosis was HGP of interest), and number of slides reviewed for each case. Missing covariates (extra renal extension and renal vein thrombosis missing on one observation) were handled using 50 iterations of chained imputations and adjusting for canonical variates.
      • vanBuuren S
      Flexible Imputation of Missing Data.
      For a frame of reference, the association between NG and OS and between necrosis grade and OS were similarly estimated adjusting for the same covariates. HGPs were classified as favorable, unfavorable or indeterminate based on each HGPs association with OS, and the extent to which the 95% confidence interval (CI) excludes effects of minimal clinical impact (indifference effects) which we defined as a positive or negative 20% change in hazard. A feature was considered favorable or unfavorable if the estimated effect was beyond minimal and the CI did not straddle both sides of the indifference zone. This allowed for considering features as favorable or unfavorable even if not statistically significant. Formally, the overlap with indifference effects makes use of the second-generation P-value.
      • Blume JD
      • D'Agostino McGowan L
      • Dupont WD
      • Greevy Jr., RA
      Second-generation p-values: Improved rigor, reproducibility, & transparency in statistical analyses.

      Risk Models

      HGP risk models, predictive of OS, were developed with the aim to stratify patients into a risk level under simple categorization rules using combinations of NG, necrosis, and HGPs. A 2-tiered model was built to classify a patient at high risk if they present themselves with NG 4, necrosis, or any unfavorable HGP; all others are classified at low risk. A 3-tiered model was built to classify a patient at high risk if they present themselves with any unfavorable HGP; at low risk if they do not qualify as high risk and present themselves with any favorable HGP and have up to 5 HGPs including no more than 2 indeterminant HGPs; and to classify all others at moderate risk. Using 1000 repeated 5-fold cross validation,
      • Friedman J HT
      • Tibshirani R
      The elements of statistical learning.
      these HGP risk models were compared to NG and necrosis grade in terms of the c-statistic and area under the receiver operating curve predicting median OS (6 years)
      • Heagerty PJ
      • Lumley T
      • Pepe MS.
      Time-dependent ROC curves for censored survival data and a diagnostic marker.
      . The association between necrosis and more than 5 HGPs and between necrosis and the number of unfavorable HGPs was tested using Fisher's Exact Test. Of note for this study, the P-values for the Kaplan-Meier curves were not cross-validated.

      Evolutionary Tree of HGPs

      To guide the development of the evolutionary tree of HGPs on OS, we noted throughout the repeated 5-fold cross validation the frequency to which each HGP was classified as favorable or unfavorable. The co-occurrence of HGPs was estimated using the Phi-coefficient of correlation
      • Guilford J.
      Pyschometric methods.
      ; and clustering of features, based on distance in correlation, were assessed in a heatmap. A combination of classification, co-occurrence, and mutual exclusion of HGPs was then used to derive an evolutionary model of tumor progression.

      Results

      Clinical and Demographic Features

      To determine the frequency of HGPs in ccRCCs, we assembled a cohort of 40 patients with non–metastatic ccRCC who survived ≥ 3 years after nephrectomy and did not progress during this time period, and 107 patients with metastatic ccRCC prior to death or loss-to-follow-up (101 metastatic within 3 years of nephrectomy). The median follow-up from the time of nephrectomy, among all cases, was 6 years. Of the 107 patients with metastatic ccRCC, 55 had non–metastatic disease, and 52 had metastatic RCC at the time of nephrectomy (de novo metastatic disease). We observed statistically significant differences in patient characteristics between patients with metastatic RCC (n = 107) versus those who did not develop metastatic disease (n = 40). Table 1 shows greater median age (60.8 vs. 56.2), tumor size (9.2 vs. 3.7 cm), increased frequency of extra-renal extension (63.6 vs. 17.5%), and renal vein thrombosis (43.0 vs. 10.0%) in metastatic versus non–metastatic cohorts (all P ≤ .001). In the metastatic group, 66% and 25%, respectively, came under the intermediate and poor IMDC categories.
      Table 1Pathologic Tumor Characteristics and Patient Demographics
      Metastatic N = 107 (%)Non–metastatic N = 40 (%)P-value
      Age at nephrectomy (median)60.856.2.001
      Gender = M79 (73.8)19 (47.5).005
      Tumor size (median)9.23.7< .001
      Extra-renal extension68 (63.6)7 (17.5)< .001
      Renal vein thrombosis46 (43.0)4 (10.0).001
      pTN stage (pTNM, AJCC 8th Edition)30
        T1aNx/N017
        T1N12
        T2Nx/N09
        T2N1210
        T3Nx/N056
        T3N115
        T46
      IMDC* categoryN/A
        Favorable14 (13.1)
        Intermediate66 (61.7)
        Poor27 (25.2)
      Nucleolar grade< .001
        230 (28.0)34 (85.0)
        354 (50.5)6 (15.0)
        423 (21.5)0 (0.0)
      Necrosis76 (71.0)10 (25.0)< .001
      Median number of HGPs (range)5.0 (2-13)4.0 (2-9).003
      Baseline covariates tested using Wilcoxon Rank Sum Test (continuous covariates) and χ2 test (categorical covariates). *IMDC- International Metsatic Disease Consortium
      As the baseline model for predicting the OS endpoint, we used the WHO/ISUP NG. Despite limited ability to predict clinical outcomes
      • Delahunt B.
      Advances and controversies in grading and staging of renal cell carcinoma.
      ,
      • Dagher J
      • Delahunt B
      • Rioux-Leclercq N
      • et al.
      Clear cell renal cell carcinoma: validation of world health organization/international society of urological pathology grading.
      ,
      • Delahunt B
      • Eble JN
      • Egevad L
      • Samaratunga H.
      Grading of renal cell carcinoma.
      ,
      • Fuhrman SA
      • Lasky LC
      • Limas C.
      Prognostic significance of morphologic parameters in renal cell carcinoma.
      , the WHO/ISUP grade is currently used for grading of ccRCC,
      • Delahunt B.
      Advances and controversies in grading and staging of renal cell carcinoma.
      ,
      • Dagher J
      • Delahunt B
      • Rioux-Leclercq N
      • et al.
      Clear cell renal cell carcinoma: validation of world health organization/international society of urological pathology grading.
      ,
      • Delahunt B
      • Cheville JC
      • Martignoni G
      • et al.
      The international society of urological pathology (ISUP) grading system for renal cell carcinoma and other prognostic parameters.
      and prediction of 5-year DFS in the UISS model
      • Zisman A
      • Pantuck AJ
      • Wieder J
      • et al.
      Risk group assessment and clinical outcome algorithm to predict the natural history of patients with surgically resected renal cell carcinoma.
      . As shown in Figure 1A, in our cohort, patients with NG2 cancers (red line) possessed a favorable OS outcome compared to patients with NG3 and NG4 cancers (blue and green lines). Adjusting for clinical predictors of OS (age, race, gender, tumor size, extrarenal extension, renal vein thrombosis), NG remained a statistically significant variable in predicting OS (P = .017). NG2 was the highest grade observed in 85% of non–metastatic cancer but surprisingly also in 28% (n = 30) of patients with mRCC (in 3-year landmark analysis: 76% of non–metastatic and 29% of metastatic patients). Necrosis, an adverse prognostic feature that has previously been incorporated into a grading model
      • Delahunt B
      • McKenney JK
      • Lohse CM
      • et al.
      A novel grading system for clear cell renal cell carcinoma incorporating tumor necrosis.
      was observed in 71% of patients with mRCC and 25% of non–metastatic RCC (Table 1; in 3-year landmark analysis: 72% of non–metastatic and 24% of mRCC patients). Adding necrosis to NG, named “necrosis grade”
      • Delahunt B
      • McKenney JK
      • Lohse CM
      • et al.
      A novel grading system for clear cell renal cell carcinoma incorporating tumor necrosis.
      resulted in a better separation of the grade groups in Kaplan Meier based analysis and was of slightly stronger statistical significance than the NG model following a cox proportional hazards model adjusted for covariates (pNG = 0.017 vs. pNecrosis Grade = 0.003) (Figure 1B).
      Figure 1
      Figure 1Grade stratification of overall survival. (A) WHO/ISUP grade (nucleolar grade); (B) Necrosis grade (WHO/ISUP grade + necrosis).

      Analysis of Histologic Growth Patterns in Primary Tumors

      Our study examined 24 HGPs that overlap in part with those previously reported (Table S1). Figure 2A shows a heatmap of architectural HGPs in 147 cases and association with metastatic status and NG. Cases in the metastatic group presented with a greater number of HGPs per tumor compared to non–metastatic cases (P < .001) and were ranked from shortest to longest OS period. Examples of HGPs are shown in Figure 2B and Figure S1. All patients displayed multiple HGPs demonstrating morphologic heterogeneity within each cancer and the number of different HGPs was inversely correlated with the length of OS (Figure 2A). In addition, oncocytic, alveolar, tubulopapillary, papillary, infiltrative, spindled low grade, sarcomatoid, fused nests, and rhabdoid HGPs were more abundant in the metastatic compared to the non–metastatic RCC group.
      Figure 2
      Figure 2Histologic Growth Patterns (HGP). (A) Heatmap of HGPs. Cases are divided into 2 groups based on metastatic state. Each column represents a case and metastatic cases are ranked based on overall survival follow-up time. (B1-12) H&E images of select HGPs. B1- small nests (100x), B2- expansile nests (100x), B3- oncocytic (400x), B4-alveolar (40x), B5- tubulopapillary (100x), B6- tubular (100x), B7- papillary (100x), B8- infiltrative (40x), B9- spindled low grade (200x), B10- sarcomatoid (200x), B11- fused nests/sheets (100x), B12- rhabdoid (200x). (C) Frequencies of HGPs in metastatic cases (blue bar, n = 107) and non–metastatic cases (orange bar, n = 40) (Color version of the figure is available online.)
      HGPs possess distinctive primary architectural and cytologic features as well as tumor associated secondary changes. The architectural patterns consist of nested/well-differentiated (small nests, expansile nests, nests with high nuclear to cytoplasmic ratio); tubulopapillary (oncocytic papillary, solid pseudopapillary, alveolar, tubulopapillary, papillary, glandular, tubular); and mesenchymal/poorly differentiated (spindled low grade, sarcomatoid, fused nests) growth patterns. Cytologic features defined additional categories of HGPs that include oncocytic and Paneth cell change, giant cells and rhabdoid differentiation. Secondary changes were also used for definition of HGPs and consisted of regressive, hemangiomatous and infiltrative growth, as well as necrosis (Figure 2B and Figure S1).
      Next, HGPs were ranked based on their frequency in the whole cohort and in the metastatic and non–metastatic subgroups (Figure S2A). Solid pseudopapillary, fused nests, alveolar, oncocytic, tubulopapillary, spindled low grade HGPs and necrosis were more frequent in metastatic cases, while small nests, regressive and cystic HGPs occurred at higher rates in non–metastatic cases (P < .05; Table S2 and S3, Figure S2A). Numbers of HGPs ranged from 2 to 9 (median 4) in non–metastatic and 2-13 (median 5) per case in metastatic cases (Figure 2C). When stratified by NG, fewer HGPs were observed in cancers with NG2 compared to cancers with NG3 or NG4 (Figure S2B) across both metastatic, and non–metastatic groups. We also observed that larger tumors contained a greater number of HGPs (Figure S2C). In short, we observed statistically significant differences in the types and frequencies of HGPs between metastatic and non–metastatic cases.

      Association of HGPs with Overall Survival

      Next, we used a Cox proportional hazards model to calculate the hazard ratio (HR) predictive of OS for each HGP (Figure 3). Based on HR and CI overlap with the indifference zone, HGPs were assigned to favorable (n = 3), indeterminate (n = 15) and unfavorable categories (n = 6). Repeated 5-fold cross-validation revealed the percent of times HGPs were classified as favorable or unfavorable HGPs in the prediction of OS time. The unfavorable category included infiltrative (P = .04), spindled low-grade (P = .01), sarcomatoid (P = .1), fused nests/sheets (P < .001), rhabdoid (P = .02) and necrosis (P = .13). These HGPs were predictive of shorter OS with HRs of 1.56-3.30. While necrosis, sarcomatoid and rhabdoid HGPs are well established negative outcomes predictors, spindled low grade, fused nests and infiltrative patterns have only recently been associated with poor OS.
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      ,
      • Cai Q
      • Christie A
      • Rajaram S
      • et al.
      Ontological analyses reveal clinically-significant clear cell renal cell carcinoma subtypes with convergent evolutionary trajectories into an aggressive type.
      The latter were identified in 42 of 107 (39.2%) of metastatic cases in our study cohort. Only one case in the non–metastatic group demonstrated an infiltrative pattern while spindled low grade, fused nests, rhabdoid, and sarcomatoid HGPs were not identified in any of the non–metastatic cases (Table S2).
      Figure 3
      Figure 3Forest plot of HGPs. A Cox regression model adjusted for age, sex, race, tumor size, extra-renal extension, renal vein thrombosis, necrosis, and number of slides reviewed per case was used to obtain the hazard ratio (HR) for overall survival of each HGP. The first column divides HGPs into favorable, indeterminate and unfavorable groups based on HR and whether confidence interval straddles minimally clinical effects (indifference zone). The second column indicates percent of cases displaying the respective HGP. The third column contains the HR and the bar in the fourth column demonstrates the confidence interval of the HR. The P-value in columns 5 is derived from the multivariate Cox model. Columns 6 and 7 list the frequency of classification of each pattern as favorable or unfavorable respectively on repeated 5-fold cross-validations.
      In contrast, small nests (P = .07), expansile nests (P = .038) and nests with high nuclear to cytoplasmic ratio (P = .09) were considered favorable with longer OS and HRs of 0.60-0.41 (Figure 3). These HGPs are present in the majority of non–metastatic cases and are also inversely correlated with the unfavorable HGPs (Figure 2A, Table S4). Overall, architectural features were most significantly associated with OS while cytologic features and secondary changes showed weaker associations.

      Overall Survival Estimation Based on Histologic Growth Patterns Risk Models

      Next, we examined how combinations of HGPs predict the length of OS after nephrectomy by generating a 2-tiered as well as a 3-tiered risk model as described in the methods section (Figure 4A, 4B) as compared to the NG and necrosis grade. The performance of the 4 models was compared in terms of their c-indices. HGP 3-tier, NG, and necrosis grade possessed a c-index of 0.73 and HGP 2-tier a c-index of 0.72 (Table 2) We compared the performance of models predicting OS at 6- years. The AUCs of HGP 3-tier and HGP 2-tier risk models amounted to 0.80 and 0.79, respectively. The AUC of necrosis grade and NG amounted to 0.78 and 0.79, respectively (Figure 4C). The presence of necrosis was associated with unfavorable HGPs (P < .001) and the number of HGPs in a given case (testing for 6+ HGPs P = .001; Figure 4C.
      Figure 4
      Figure 4HGP based risk models. (A) 2-tiered risk model. Cases in the high-risk group (blue line) contain either unfavorable HGP, necrosis or NG4. Cases in the low-risk group (red line) lack these features. (B) 3-tiered risk model. Cases are separated into 3 groups of low (HGP-RS1), moderate (HGP-RS2) and high risk (HGP-RS3). HGP-RS1 is defined by at least 1 favorable HGPs and not more than 2 indeterminate HGPs; HGP-RS2 consists of cases with more than 2 indeterminate HGPs and no unfavorable HGP; HGP-RS3 cases contain any of unfavorable HGPs. (C) c-statistic and AUC box plots for nucleolar grade, necrosis grade, 2 and 3-tiered risk models following repeated 5-fold cross-validation; (D) Box plot demonstrating the association between necrosis and the frequency of HGPs per case (Color version of the figure is available online.)
      Table 2Reclassification of WHO/ISUP Grade (Nucleolar Grade)
      Nucleolar GradeNecrosis Grade3-Tier HGP
      Nucleolar GradeG2G3G4RS1RS2RS3
      NG240240312211
      NG301743162519
      NG400234415
      Cross-validated c-statistic0.730.730.73
      Cross-validated 6-y AUC0.780.790.80
      Altogether, these results suggest that morphologic tumor heterogeneity may be as informative as NG alone or in combination with necrosis in predicting OS after nephrectomy.

      Evolution of Histologic Growth Patterns

      We hypothesized that progression from low-grade to high-grade in ccRCC is related to the evolution of HGPs and may define a tumor evolutionary tree. This prompted a calculation of pairwise correlations between HGPs. Figure 5A depicts the pairwise correlation matrix of HGPs. As expected, the highest correlations were observed between NG4 and sarcomatoid or rhabdoid with correlation coefficients of 0.56 and 0.48, respectively (Table S4). In addition, modest correlations were observed between tubulopapillary, solid pseudopapillary, tubular and NG3 (rho = 0.19-0.27) suggesting that tubular and papillary morphologies are associated with higher NG. Modest correlations also occurred between fused nests/sheets, low grade spindled, sarcomatoid and rhabdoid (rho = 0.17-0.27) and between small and expansile nests (rho = 0.28). These results suggest 2 evolutionary branches (Figure 5B) in ccRCC, one characterized by mesenchymal differentiation and the other by epithelial tubulopapillary morphology. Mesenchymal and tubulopapillary HGPs were rarely observed in the same cancer. The mesenchymal HGPs possess HRs of 2.3-3.3, while the highest HR in tubulopapillary HGPs is 1.4 for tubular HGP and, with an unfavorable classification that accounts for the 20% indifference zone, tubular HGP was identified as an unfavorable HGP in 30% of repeated 5-fold cross-validation models. Correlations were observed with NG4 (positive) and NG2 (negative) in mesenchymal and epithelial branches, respectively. Of 107 metastatic cases, 44 exhibited mesenchymal differentiation, while only 1 case with mesenchymal differentiation was detected in the non–metastatic group. While mesenchymal differentiation is considered the main morphologic predictor of metastatic risk only 44 of 107 metastatic cases displayed mesenchymal differentiation. In the other cases without mesenchymal differentiation, metastatic cells may exist in areas of epithelial, and tubulopapillary HGPs. In summary, the correlation between HGPs, tumor invasion and reduced OS connects tumor cell morphology and behavior with tumor cell aggression.
      Figure 5
      Figure 5Tumor evolution. (A) Pairwise correlation matrix. Spearman's correlation coefficients of paired HGPs. Correlation coefficients greater than 0.25 or less than - 0.25 are depicted by circles within colored squares. Since the color bar ranges from - 0.25 to + 0.25, correlation coefficients greater than 0.25 or less than – 0,25 are not separated by color intensity. Refer to Table S4 for coefficient values. (B) Evolutionary diagram of HGPs. 1, 2, and 3 indicate different evolutionary pathways. Pathway 1 is characterized by mesenchymal differentiation, pathway 2 by tubulo-papillary architecture and pathway 3 by a nested morphology.

      Discussion

      In this study, a systematic analysis of HGPs demonstrates significant morphologic heterogeneity across ccRCCs. Specific HGPs (fused nests/sheets, spindled low grade, rhabdoid, sarcomatoid, and infiltrative behavior) were associated with decreased OS post nephrectomy while others (small nests, expansile nests, and nests with high nuclear to cytoplasmic ratio) predicted longer OS. The results from this study and 2 previous studies demonstrating the prognostic value of HGPs
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      ,
      • Cai Q
      • Christie A
      • Rajaram S
      • et al.
      Ontological analyses reveal clinically-significant clear cell renal cell carcinoma subtypes with convergent evolutionary trajectories into an aggressive type.
      suggest that HGPs are associated with tumor grade, DFS, and OS post nephrectomy. Verine et al. propose to replace the conventional WHO/ISUP grade with a new, 4-tiered HGP based grading system in which each cancer is graded based on the HGP with the highest likelihood ratio for risk of cancer recurrence.
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      In contrast, Cai et al. developed a grading nomogram for prediction of DFS that includes both favorable (tubular/acinar) as well as unfavorable (chromophobe-like, infiltration) HGPs
      • Cai Q
      • Christie A
      • Rajaram S
      • et al.
      Ontological analyses reveal clinically-significant clear cell renal cell carcinoma subtypes with convergent evolutionary trajectories into an aggressive type.
      in addition to the WHO/ISUP grade and necrosis. Our study confirms salient findings from these 2 studies: (i) unfavorable HGPs are observed in the metastatic group and not in non–metastatic cases; (ii) the number and type of HGPs are significantly associated with OS after adjusting for other clinical variables; (iii) morphologic intra-tumoral heterogeneity is associated with grade and disease progression.
      In addition to providing independent evidence for a role of HGPs in the estimation of OS, our study differs from previous studies in the following ways: (i) HGPs were defined before the other publications became available and therefore consist of 14 overlapping HGPs and 10 non–/partially overlapping HGPs (Table S1); (ii) we compared the impact of NG, necrosis and HGPs in 4 models predicting OS in ccRCCs; (iii) we combined pairwise correlations between HGPs and HRs of HGPs to develop novel insights into tumor evolution. Altogether, our study design, which includes a proportionally large number of metastatic cases allowed us to understand the relationship between HGPs, tumor heterogeneity, necrosis, NG, and OS in patients undergoing nephrectomy for ccRCC.
      Although several genomic studies for ccRCC have demonstrated tumor heterogeneity and tumor evolution, these have been performed agnostic to the HGPs and have instead focused only on the subtypes of RCC or percentage of clear cells.
      • Turajlic S
      • Xu H
      • Litchfield K
      • et al.
      Tracking cancer evolution reveals constrained routes to metastases: TRACERx renal.
      • Turajlic S
      • Xu H
      • Litchfield K
      • et al.
      Deterministic evolutionary trajectories influence primary tumor growth: TRACERx renal.
      • Gerlinger M
      • Horswell S
      • Larkin J
      • et al.
      Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing.
      • Gerlinger M
      • Rowan AJ
      • Horswell S
      • et al.
      Intratumor heterogeneity and branched evolution revealed by multiregion sequencing.
      • Ricketts CJ
      • De Cubas AA
      • Fan H
      • et al.
      The cancer genome atlas comprehensive molecular characterization of renal cell carcinoma.
      Cancer genome atlas research n. comprehensive molecular characterization of clear cell renal cell carcinoma.
      Multiregion sequencing of ccRCCs identified mutational heterogeneity and spatially separate subclones with driver alterations and correlation between number of driver alterations and sampled regions.
      • Gerlinger M
      • Horswell S
      • Larkin J
      • et al.
      Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing.
      Furthermore, subclonal driver mutations in molecularly distinct oncogenic pathways within the same tumor were used to reconstruct a map of genomic tumor evolution. In the future, it will be interesting to determine the connections between morphologic, and genomic heterogeneity. The histologic heterogeneity may also be relevant in determining treatment response. Although largely limited in inclusion of HGPs in patient assignment, results from clinical trials such as the recent keynote 564 trial that included localized ccRCCs with adverse sarcomatoid HGP showed improved disease-free survival with pembrolizumab compared to placebo group.
      • Choueiri TK
      • Tomczak P
      • Park SH
      • et al.
      Adjuvant pembrolizumab after nephrectomy in renal-cell carcinoma.
      In the future, it may be interesting to evaluate if other HGPs, and/or evolutionary pathways are predictive of response to systemic therapy.
      Even though we used the OS and not DFS endpoint as used by Cai et al., evolutionary trajectories of both cohorts are remarkably similar. We arrived at similar conclusions as Cai et al. by using an aggregate of HR and correlation coefficients and without the elegant modeling performed in the Cai et al. paper. The large number of tumors with rhabdoid and sarcomatoid HGPs in our cohort led to an expansion of the evolutionary tree and the appearance of mesenchymal HGPs emerging from fused nests, which corresponds to the sheet-like growth patter used by Cai et al. We agree with Cai et al. on 2 distinct morphologic evolutionary branches (mesenchymal and tubulopapillary epithelial) and on the observation that the alveolar HGP may represent a dead-end differentiation state without metastatic progression (HR = 0.9).
      Despite the rigor in the statistical data analysis, this study has several limitations: (i) the cohort includes a high number of metastatic cases and does not represent a population-based cohort of ccRCCs. Because of the high number of metastatic cases, there is an adequate representation of unfavorable HGPs, however, it precludes us from using DFS as an endpoint. Metastasis at time of nephrectomy was present in 48.6% of metastatic cases reducing the cases for DFS assessment by 35% (52 of 147). Despite these limitations, the conclusions from our study are in line with the 2 previous studies that used a population based cohort; (ii) evaluation of HGPs was conducted by a single genitourinary -subspecialty trained pathologist using retrospective cases, however good interobserver concordance for morphologic assessment has been reported across studies by another group
      • Verine J
      • Colin D
      • Nheb M
      • et al.
      Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
      ; (iii) we are working under the premise that the HGPs are variant patterns seen in ccRCCs and do not represent collision tumors of non–clear cell morphologies that arise separately from the proximal convoluted tubules
      • Delahunt B
      • Eble JN
      • Egevad L
      • Samaratunga H.
      Grading of renal cell carcinoma.
      ; (iv) the evolutionary analysis of HGP is based on correlation coefficients and does not conclusively demonstrate the directionality of the transition from one growth pattern to another; (v) the study did not include drug treatment data. Future studies that include spatially resolved genomic and proteomic data will be needed to map out the exact path of tumor evolution in ccRCC.

      Conclusion

      Our data support the novel concept that HGPs in the primary ccRCC can be used to estimate OS after nephrectomy. HGPs can be grouped into favorable, indeterminate, and unfavorable categories based on their HR to predict OS. The evolutionary analysis of HGPs strongly suggests 2 divergent branches, one that contains the mesenchymal HGPs, and the other that consists of tubulopapillary epithelial HGPs which have not been studied with regard to their metastatic genomic pathways. Altogether, the study demonstrates that simple combinations of HGPs can be used to estimate OS after nephrectomy and should be evaluated further for inclusion in prognostic models.

      Clinical Practice Points

      Histologic growth patterns (HGPs) in clear cell renal cell carcinoma have limited clinical significance, and currently are included only in defining grade 4 tumors, with majority of these tumors designated as grade 2, and 3. There exists significant overlap in survival outcomes across these intermediate grade tumors. Recently 2 separate studies have shown promising associations of HGPs with survival outcomes. In the current study we investigated the clinical significance of 24 HGPs in predicting survival outcomes across 147 primary nephrectomy specimens, that included 107 metastatic, and 40 non–metastatic cases. In addition to the well-established rhabdoid and sarcomatoid histology that are associated with adverse outcomes, we identified 2 additional mesenchymal patterns of low grade spindled and fused nests that also conferred adverse prognosis. Furthermore, the co-occurrence of HGPs and hazard ratios for overall survival revealed 2 separate evolutionary pathways that differed in their metastatic potential: one with mesenchymal differentiation and other with epithelial tubulopapilalry differentiation. Findings from this study suggest that HGPs may provide a novel path to refine the estimation of OS after nephrectomy and to determine the molecular basis of tumor evolution.

      Author Contributions

      Deepika Sirohi: Pathology review, pathology data compilation, study design, conceptualization, manuscript writing; Beatrice Knudsen: Study design, conceptualization, manuscript writing; Jonathan Chipman: Statistical analysis, manuscript writing; Ben Haaland: Review of statistical methods, manuscript review and edits; Evan Raps: Pathology review, pathology data compilation, manuscript review and edits; Marc Barry, Dan Albertson, Jon Mahlow, Ting Liu: pathology consultations, manuscript review and edits; Nicolas Sayegh; Haoran Li; Nityam Rathi; Prayushi Sharma; Neeraj Agarwal: clinical data compilation, manuscript review, and edits. Conflicts of Interest: Deepika Sirohi: Consultancy: Genentech. Ben Haaland: Consultancy: Astra Zeneca, Prometics Life Sciences, Value Analytics, and the National Kidney Foundation. Funding: travel funds from Flatiron HealthNeeraj Agarwal: Consultancy: Astellas, Astra Zeneca, Aveo, Bayer, Bristol Myers Squibb, Calithera, Clovis, Eisai, Eli Lilly, EMD Serono, Exelixis, Foundation Medicine, Genentech, Janssen, Merck, MEI Pharma, Nektar, Novartis, Pfizer, Pharmacyclics, and Seattle Genetics. Research funding institution: Astra Zeneca, Bavarian Nordic , Bayer, Bristol Myers Squibb, Calithera, Celldex, Clovis, Eisai, Eli Lilly, EMD Serono, Exelixis, Genentech, Glaxo Smith Kline, Immunomedics, Janssen, Medivation, Merck, Nektar, New Link Genetics, Novartis, Pfizer, Prometheus, Rexahn, Roche, Sanofi, Seattle Genetics, Takeda, and Tracon.

      Disclosures/Acknowledgments

      The study was presented as an abstract at USCAP 2021 and has been accepted for abstract presentation at Kidney Cancer Research Summit 2021.

      Acknowledgments

      We would like to acknowledge the support from the Computational Oncology Research initiative , Genitourinary Cancer Program at the Huntsman Cancer Institute and the Department of Pathology , University of Utah and ARUP Laboratories .

      Funding

      We would like to acknowledge the support from the Computational Oncology Research initiative , Genitourinary Cancer Program at the Huntsman Cancer Institute and the Department of Pathology , University of Utah and ARUP Laboratories .

      Appendix. Supplementary materials

      References

      1. American Cancer Society. Cancer Statistics Center. (Accessed at: June 30, 2021 Accessed from: https://cancerstatisticscenter.cancer.org/?_ga=2.180760999.1670122945.1625274963-472095398.1625274963#!/.)

        • Siegel RL
        • Miller KD
        • Jemal A.
        Cancer statistics.
        CA Cancer J Clin. 2018; 2018: 7-30
        • Dabestani S
        • Thorstenson A
        • Lindblad P
        • Harmenberg U
        • Ljungberg B
        • Lundstam S.
        Renal cell carcinoma recurrences and metastases in primary non-metastatic patients: a population-based study.
        World J Urol. 2016; 34: 1081-1086
        • Rabinovitch RA
        • Zelefsky MJ
        • Gaynor JJ
        • Fuks Z.
        Patterns of failure following surgical resection of renal cell carcinoma: implications for adjuvant local and systemic therapy.
        J Clin Oncol. 1994; 12: 206-212
        • Motzer RJ
        • Tannir NM
        • McDermott DF
        • et al.
        Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma.
        N Engl J Med. 2018; 378: 1277-1290
        • Zisman A
        • Pantuck AJ
        • Wieder J
        • et al.
        Risk group assessment and clinical outcome algorithm to predict the natural history of patients with surgically resected renal cell carcinoma.
        J Clin Oncol. 2002; 20: 4559-4566
        • Delahunt B.
        Advances and controversies in grading and staging of renal cell carcinoma.
        Mod Pathol. 2009; 22: S24-S36
        • Delahunt B
        • McKenney JK
        • Lohse CM
        • et al.
        A novel grading system for clear cell renal cell carcinoma incorporating tumor necrosis.
        Am J Surg Pathol. 2013; 37: 311-322
        • Dagher J
        • Delahunt B
        • Rioux-Leclercq N
        • et al.
        Clear cell renal cell carcinoma: validation of world health organization/international society of urological pathology grading.
        Histopathology. 2017; 71: 918-925
        • Delahunt B
        • Cheville JC
        • Martignoni G
        • et al.
        The international society of urological pathology (ISUP) grading system for renal cell carcinoma and other prognostic parameters.
        Am J Surg Pathol. 2013; 37: 1490-1504
        • Moch HHP
        • Ulbright TM
        • Reuter VE
        WHO classification of tumors of the urinary system and male genital organs.
        4th ed. IARC Press, Lyon2016
        • Khor LY
        • Dhakal HP
        • Jia X
        • et al.
        Tumor necrosis adds prognostically significant information to grade in clear cell renal cell carcinoma: a study of 842 consecutive cases from a single institution.
        Am J Surg Pathol. 2016; 40: 1224-1231
        • Kuthi L
        • Jenei A
        • Hajdu A
        • et al.
        Prognostic factors for renal cell carcinoma subtypes diagnosed according to the 2016 WHO renal tumor classification: a study involving 928 patients.
        Pathol Oncol Res. 2017; 23: 689-698
        • Kryvenko ON
        • Roquero L
        • Gupta NS
        • Lee MW
        • Epstein JI.
        Low-grade clear cell renal cell carcinoma mimicking hemangioma of the kidney: a series of 4 cases.
        Arch Pathol Lab Med. 2013; 137: 251-254
        • Williamson SR
        • MacLennan GT
        • Lopez-Beltran A
        • et al.
        Cystic partially regressed clear cell renal cell carcinoma: a potential mimic of multilocular cystic renal cell carcinoma.
        Histopathology. 2013; 63: 767-779
        • Montironi R
        • Lopez-Beltran A
        • Cheng L
        • Galosi AB
        • Montorsi F
        • Scarpelli M.
        Clear cell renal cell carcinoma (ccRCC) with hemangioblastoma-like features: a previously unreported pattern of ccRCC with possible clinical significance.
        Eur Urol. 2014; 66: 806-810
        • Dagher J
        • Delahunt B
        • Rioux-Leclercq N
        • et al.
        Assessment of tumour-associated necrosis provides prognostic information additional to World Health Organization/International Society of Urological Pathology grading for clear cell renal cell carcinoma.
        Histopathology. 2019; 74: 284-290
        • Verine J
        • Colin D
        • Nheb M
        • et al.
        Architectural patterns are a relevant morphologic grading system for clear cell renal cell carcinoma prognosis assessment: comparisons with WHO/ISUP grade and integrated staging systems.
        Am J Surg Pathol. 2018; 42: 423-441
        • Cai Q
        • Christie A
        • Rajaram S
        • et al.
        Ontological analyses reveal clinically-significant clear cell renal cell carcinoma subtypes with convergent evolutionary trajectories into an aggressive type.
        EBioMedicine. 2020; 51102526
        • Kapur P
        • Christie Alana
        • Rajaram Satwik
        • et al.
        What morphology can teach us about renal cell carcinoma clonal evolution.
        Kidney Cancer Journal. 2020; 18: 68-76
        • Barnarad G.
        A new test for 2 x 2 tables.
        Nature. 1945; 156: 177
        • vanBuuren S
        Flexible Imputation of Missing Data.
        CRC Press/Taylor and Francis Group, Boca Raton, FL2012
        • Blume JD
        • D'Agostino McGowan L
        • Dupont WD
        • Greevy Jr., RA
        Second-generation p-values: Improved rigor, reproducibility, & transparency in statistical analyses.
        PLoS One. 2018; 13e0188299
        • Friedman J HT
        • Tibshirani R
        The elements of statistical learning.
        Springer, New York, NY2001 (New York: Springer series in statistics)
        • Heagerty PJ
        • Lumley T
        • Pepe MS.
        Time-dependent ROC curves for censored survival data and a diagnostic marker.
        Biometrics. 2000; 56: 337-344
        • Guilford J.
        Pyschometric methods.
        McGraw-Hill Book Company, New York, NY1954
        • Delahunt B
        • Eble JN
        • Egevad L
        • Samaratunga H.
        Grading of renal cell carcinoma.
        Histopathology. 2019; 74: 4-17
        • Fuhrman SA
        • Lasky LC
        • Limas C.
        Prognostic significance of morphologic parameters in renal cell carcinoma.
        Am J Surg Pathol. 1982; 6: 655-663
        • Turajlic S
        • Xu H
        • Litchfield K
        • et al.
        Tracking cancer evolution reveals constrained routes to metastases: TRACERx renal.
        Cell. 2018; 173: e12
        • Turajlic S
        • Xu H
        • Litchfield K
        • et al.
        Deterministic evolutionary trajectories influence primary tumor growth: TRACERx renal.
        Cell. 2018; 173: e11
        • Gerlinger M
        • Horswell S
        • Larkin J
        • et al.
        Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing.
        Nat Genet. 2014; 46: 225-233
        • Gerlinger M
        • Rowan AJ
        • Horswell S
        • et al.
        Intratumor heterogeneity and branched evolution revealed by multiregion sequencing.
        N Engl J Med. 2012; 366: 883-892
        • Ricketts CJ
        • De Cubas AA
        • Fan H
        • et al.
        The cancer genome atlas comprehensive molecular characterization of renal cell carcinoma.
        Cell Rep. 2018; 23: 3698
      2. Cancer genome atlas research n. comprehensive molecular characterization of clear cell renal cell carcinoma.
        Nature. 2013; 499: 43-49
        • Choueiri TK
        • Tomczak P
        • Park SH
        • et al.
        Adjuvant pembrolizumab after nephrectomy in renal-cell carcinoma.
        N Engl J Med. 2021; 385: 683-694