6. PROGNOSTIC FACTORS
6.1. Classification
Prognostic factors can be classified into: anatomical, histological, clinical, and molecular.
6.2. Anatomical factors
Tumour size, venous invasion and extension, collecting system invasion, perinephric- and sinus fat invasion, adrenal involvement, and LN and distant metastasis are included in the TNM classification system [182,183] (Table 4.1).
6.3. Histological factors
Histological factors include tumour grade, RCC subtype, lymphovascular invasion, tumour necrosis, and invasion of the collecting system [184,185]. Tumour grade is considered one of the most important histological prognostic factors. Fuhrman nuclear grade [186] has now been replaced by the WHO/ISUP grading classification [187]. This relies solely on nucleolar prominence for grade 1-3 tumours, allowing for less inter-observer variation [188]. It has been shown that the WHO/ISUP grading provides superior prognostic information compared to Fuhrman grading, especially for grade 2 and grade 3 tumours [189]. Rhabdoid and sarcomatoid changes can be found in all RCC types and are equivalent to grade 4 tumours. Sarcomatoid changes are more often found in chRCC than other subtypes [190]. The percentage of the sarcomatoid component appears to be prognostic as well, with a larger percentage of involvement being associated with worse survival. There is no agreement on the optimal prognostic cut-off for sub-classifying sarcomatoid changes [191,192], although 20% has been suggested to distinguish focal and extensive amount of sarcomatoid features [193]. The WHO/ISUP grading system is applicable to both ccRCC and pRCC. It is currently not recommended to grade chRCC. However, a recent study suggested a two-tiered chRCC grading system (low vs. high grade) based on the presence of sarcomatoid differentiation and/or tumour necrosis, which was statistically significant on multivariable analysis [194]. Both the WHO/ISUP and chRCC grading systems need to be validated for prognostic systems and nomograms [187].
Renal cell carcinoma subtype is regarded as another important prognostic factor. On univariable analysis, patients with chRCC vs. pRCC vs. ccRCC had a better prognosis [195,196] (Table 6.1). However, prognostic information provided by the RCC type is lost when stratified according to tumour stage [196,197] (LE: 3).
In a recent cohort study of 1,943 patients with ccRCC and pRCC significant survival differences were only shown between pRCC type I and ccRCC [198]. Papillary RCC has been traditionally divided into type 1 and 2, but a subset of tumours shows mixed features. For more details, see Section 3.2 – Histological diagnosis. Data also suggest that type 2 pRCC is a heterogeneous entity with multiple molecular subgroups [199]. Some studies suggest poorer survival for type 2 than type 1 [200], but this association is often lost in the multivariable analysis [201]. A meta-analysis did not show a significant survival difference between both types [202,203].
TFE3 re-arranged RCC (formerly called RCC with Xp11.2 translocation) has a poor prognosis [204]. Its incidence is low, but its presence should be systematically assessed in young patients. Renal cell carcinoma type classification has been confirmed by cytogenetic and genetic analyses [205-207] (LE: 2b). Surgically excised malignant complex cystic masses contain ccRCC in the majority of cases, and more than 80% are pT1. In a recent series, 5-year CSS was 98% [208]. Differences in tumour stage, grade and CSS between RCC types are illustrated in Table 6.1.
Table 6.1: Baseline characteristics and cancer-specific survival of surgically treated patients by RCC type [145]
Survival time | % RCC | % Sarcomatoid | % T3-4 | % N1 | % M1 | % 10 year CSS (%) |
Clear-cell RCC | 80 | 5 | 33 | 5 | 15 | 62 |
Papillary RCC | 15 | 1 | 11 | 5 | 3 | 86 |
Chromophobe RCC | 5 | 8 | 15 | 4 | 4 | 86 |
CSS = cancer-specific survival.
In all RCC types, prognosis worsens with stage and histopathological grade (Table 6.2). The 5-year OS for all types of RCC is 49%, which has improved since 2006, probably due to an increase in incidentally detected RCCs and new systemic treatments [209,210]. Although not considered in the current N classification, the number of metastatic regional LNs is an important predictor of survival in patients without distant metastases [211].
Table 6.2: Cancer-specific survival by stage [212]
Grade | HR (95% CI) |
T1N0M0 | Referent |
T2N0M0 | 2.71 (2.17–3.39) |
T3N0M0 | 5.20 (4.36–6.21) |
T4N0M0 | 16.88 (12.40–22.98) |
N+M0 | 16.33 (12.89–20.73) |
M+ | 33.23 (28.18–39.18) |
CI = confidence interval; HR = hazard ratio.
6.4. Clinical factors
Clinical factors include performance status (PS), local symptoms, cachexia, anaemia, platelet count, neutrophil count, lymphocyte count, C-reactive protein (CRP) [213], albumin, and various indices deriving from these factors such as the neutrophil-to-lymphocyte ratio (NLR) [103,214-219] (LE: 3). As a marker of systemic inflammatory response, a high pre-operative NLR has been associated with poor prognosis [220], but there is significant heterogeneity in the data and no agreement on the optimal prognostic cut-off. Even though obesity is an aetiological factor for RCC, it has also been observed to provide prognostic information. A high body mass index (BMI) appears to be associated with improved survival outcomes in both non-metastatic and metastatic RCC [221-223]. This association is linear with regards to cancer-specific mortality (CSM), while obese RCC patients show increasing all-cause mortality with increasing BMI [224]. There is also evolving evidence on the prognostic value of body composition indices measured on cross-sectional imaging, such as sarcopenia and fat accumulation [219,225,226].
6.5. Molecular factors
Numerous molecular markers such as carbonic anhydrase IX (CaIX), VEGF, HIF, Ki67 (proliferation), p53, p21 [227], PTEN (phosphatase and tensin homolog) cell cycle [228], E-cadherin, osteopontin [229] CD44 (cell adhesion) [230,231], CXCR4 [232], PD-L1 [233], miRNA, SNPs, gene mutations, and gene methylations have been investigated (LE: 3) [234]. While the majority of these markers are associated with prognosis and many improve the discrimination of current prognostic models, there has been very little emphasis on external validation studies. Furthermore, there is no conclusive evidence on the value of molecular markers for treatment selection in mRCC [213,233,235]. Their routine use in clinical practice is therefore not recommended.
Several prognostic and predictive marker signatures have been described for specific systemic treatments in mRCC. In the JAVELIN Renal 101 trial (NCT02684006), a 26-gene immunomodulatory gene signature predicted PFS in those treated with avelumab plus axitinib, while an angiogenesis gene signature was associated with PFS for sunitinib. Mutational profiles and histocompatibility leukocyte antigen (HLA) types were also associated with PFS, while programmed death-ligand 1 (PD-L1) expression and tumour mutational burden were not [236]. In IMmotion151 (NCT02420821), a T effector/IFN-γ-high or angiogenesis-low gene expression signature predicted improved PFS for atezolizumab plus bevacizumab compared to sunitinib. The angiogenesis-high gene expression signature correlated with longer PFS in patients treated with sunitinib [237]. In CheckMate 214 (NCT02231749), a higher angiogenesis gene signature score was associated with better overall response rates and PFS for sunitinib, while a lower angiogenesis score was associated with higher ORR in those treated with nivolumab plus ipilimumab. Progression-free survival > 18 months was more often seen in patients with higher expression of Hallmark inflammatory response and Hallmark epithelial mesenchymal transition gene sets [219].
Urinary and plasma Kidney-Injury Molecule-1 (KIM-1) has been identified as a potential diagnostic and prognostic marker. KIM-1 concentrations were found to predict RCC up to five years prior to diagnosis and were associated with a shorter survival time [238]. KIM-1 is a glycoprotein marker of acute proximal tubular injury and therefore mainly expressed in RCC derived from the proximal tubules such as ccRCC and pRCC [239]. While early studies are promising, more high-quality research is required. Several retrospective studies and large molecular screening programmes have identified mutated genes and chromosomal changes in ccRCC with distinct clinical outcomes. The expression of the BAP1 and PBRM1 genes, situated on chromosome 3p in a region that is deleted in more than 90% of ccRCCs, have shown to be independent prognostic factors for tumour recurrence [240-242]. These published reports suggest that patients with BAP1- mutant tumours have worse outcomes compared with patients with PBRM1-mutant tumours [241]. Loss of chromosome 9p and 14q have been consistently shown to be associated with poorer survival [243-245]. The TRACERx renal consortium has proposed a genetic classification based on RCC evolution (punctuated vs. branched vs. linear), which correlates with tumour aggressiveness and survival [244]. Additionally, a 16-gene signature was shown to predict disease-free survival (DFS) in patients with non-metastatic RCC [246]. However, these signatures have not been validated by independent researchers yet.
6.6. Prognostic models
Prognostic models combining independent prognostic factors have been developed and externally validated [247-254]. These models are more accurate than TNM stage or grade alone for predicting clinically relevant oncological outcomes (LE: 3). Before being adopted, new prognostic models should be evaluated and compared to current prognostic models with regards to discrimination, calibration and net benefit. Pathological prognostic factors are used in the Leibovich 2003 score/groups for clear cell [250]. There are prognostic models for non-clear cell RCC, like the VENUSS score for papillary RCC [201]. Although both were validated in several studies and showed superior discrimination to other prognostic models, molecular markers are needed [255-258]. In metastatic disease, risk groups assigned by the Memorial Sloan Kettering Cancer Centre (MSKCC) (primarily created in the pre-targeted therapy era and validated in patients receiving targeted therapy) and the IMDC (initially created in the targeted therapy era) differ in 23% of cases [259]. The IMDC model has been used in most of the recent RCTs, including those with immune checkpoint inhibitors (ICIs), and may therefore be the preferred model for clinical practice. The discrimination of the IMDC models improves by additional variables, such as presence of brain metastasis, bone metastasis, liver metastasis, NLR and platelet count [260-263]. Tables 6.3 and 6.4 summarise the current most relevant prognostic models.
6.7. Summary of evidence and recommendations for prognostic factors
Summary of evidence | LE |
In RCC patients, TNM stage, tumour size, grade, and RCC subtype provide important prognostic information. | 2a |
The 2003 Leibovich score is a validated prognostic model to predict the short- and long-term risk of metastasis in individual patients with sporadic, unilateral pT1-4 N0/+ M0 clear cell renal cell carcinoma. | 2b |
The VENUSS score is a validated prognostic model to predict the short- and long-term risk of disease recurrence in individual patients with sporadic, unilateral pT1-4 N0/+ M0 papillary renal cell carcinoma. | 2b |
Recommendations | Strength rating |
Use the current Tumour, Node, Metastasis classification system. | Strong |
Use the WHO/ISUP grading system and classify renal cell carcinoma type. | Strong |
Use prognostic models in localised and metastatic disease. | Strong |
Use the 2003 Leibovich scoring model for risk stratification of localised and locally advanced clear cell renal cell carcinoma. | Weak |
Use the VENUSS scoring model for risk stratification of localised and locally advanced papillary renal cell carcinoma. | Weak |
Do not routinely use molecular markers to assess prognosis. | Strong |
Table 6.3: Prognostic models for localised RCC
Prognostic model | Subtype* | Risk factors/prognostic factors |
UISS** [264] | All | 1. ECOG PS 2. T classification 3. N classification (N+ classified as metastatic) 4. Grade T1N0M0G1-2, ECOG PS 0: low-risk disease T3N0M0G2-4, ECOG PS ≥ 1 OR T4N0M0: high-risk disease Any other N0M0: intermediate-risk disease |
Leibovich score/model 2003 [250] | CC | 1. T classification (pT1a: 0, pT1b: 1, pT2:3, pT3-4: 4 points) 2. N classification (pNx/N0: 0, pN+: 2 points) 3. Tumour size (< 10 cm: 0, ≥ 10 cm: 1 point) 4. Grade (G1-2: 0, G3: 1, G4: 3 points) 5. Tumour necrosis (absent: 0, present: 1 point) 0-2 points: low-risk disease 3-5 points: intermediate-risk disease 6 or more points: high-risk disease |
Leibovich score/model 2018 [265] | CC, P, CH | ccRCC Progression (9 factors): constitutional symptoms, grade, tumour necrosis, sarcomatoid features, tumour size, perinephric or sinus fat invasion, tumour thrombus level, extension beyond kidney, nodal involvement. Cancer-specific survival (12 factors): age, ECOG PS, constitutional symptoms, adrenalectomy, surgical margins, grade, tumour necrosis, sarcomatoid features, tumour size, perinephric or sinus fat invasion, tumour thrombus, nodal involvement. No risk groups /prognostic groups. pRCC Low risk (group 1): grade 1-2, no fat invasion, no tumour thrombus. Intermediate risk (group 2): grade 3, no fat invasion, no tumour thrombus. High risk (group 3): grade 4 or fat invasion or any level tumour thrombus. chRCC Low risk (group 1): no fat invasion, no sarcomatoid differentiation, no nodal involvement. Intermediate risk (group 2): fat invasion and no sarcomatoid differentiation and no nodal involvement. High risk (group 3): sarcomatoid differentiation or nodal involvement. |
P | 1. T classification (pT1: 0, pT2: 1, pT3-4: 2 points) 2. N classification (pNx/pN0: 0, pN1: 3 points) 3. Tumour size (≤ 4 cm: 0, > 4 cm: 2 points) 4. Grade (G1/2: 0, G3/4: 2 points) 5. Tumour thrombus (absent: 0, present: 2 points) 0-2 points: low-risk disease 3-5 points: intermediate-risk disease 6 or more points: high-risk disease | |
GRANT score/model**** [266] | All | 1. Age > 60 years 2. T classification = T3b, pT3c or pT4 3. N classification = pN1 4. (Fuhrman) grade = G3 or G4 0-1 factors: favourable-risk disease 2 or more factors: unfavourable-risk disease |
* ccRCC = clear-cell RCC; ECOG = Eastern Cooperative Oncology Group; pRCC = papillary RCC; chRCC = chromophobe RCC; PS = performance status.
** University of California Integrated Staging system. Available at https://www.mdcalc.com/ucla-integratedstaging-system-uiss-renal-cell-carcinoma-rcc.
*** Venous extension, Nuclear grade, Size, Stage. Available at https://www.evidencio.com/models/show/2369.
**** Grade, Age, Nodes and Tumour.
Table 6.4: Prognostic models for metastatic RCC
Prognostic model | Subtype | Risk factors/prognostic factors |
MSKCC [267]** | All | 1. Karnofsky PS [268]* < 80% 2. Interval from diagnosis to systemic treatment < 1 year 3. Haemoglobin < Lower Limit of Normal 4. Corrected calcium >10mg/dL/> 2.5 mmol/L 5. LDH > 1.5x Upper Limit of Normal 0 factors: favourable-risk disease 1-2 factors: intermediate-risk disease 3-5 factors: poor-risk disease |
IMDC [269]*** | All | 1. Karnofsky PS [268]* < 80% 2. Interval from diagnosis to treatment < 1 year 3. Haemoglobin < lower limit of normal 4. Corrected calcium > upper limit of normal (i.e. > 10.2 mg/dL) 5. Neutrophil count > upper limit of normal (i.e. > 7.0×10⁹/L) 6. Platelet count > upper limit of normal (i.e. > 400,000) 0 factors: favourable-risk disease 1-2 factors: intermediate-risk disease 3-6 factors: poor-risk disease |
IMDC = International Metastatic Renal Cancer Database Consortium; LDH = lactate dehydrogenase;MSKCC = Memorial Sloan Kettering Cancer Center; PS = performance status.
* Karnofsky performance status calculator: https://www.thecalculator.co/health/Karnofsky-Score-for-Performance-Status-Calculator-961.html.
** MSKCC: https://www.mdcalc.com/memorial-sloan-kettering-cancer-center-mskcc-motzer-score-etastaticrenal-cell-carcinoma-rcc.
*** IMDC: https://www.mdcalc.com/imdc-international-metastatic-rcc-database-consortium-risk-score-rcc.