Memorial Sloan-Kettering Cancer Center Model Research Articles

Population-Based Validation of the MIA and MSKCC Tools for Predicting Sentinel Lymph Node Status

Patients with melanoma are selected for sentinel lymph node biopsy (SLNB) based on their risk of a positive SLN. To improve selection, the Memorial Sloan Kettering Cancer Center (MSKCC) and Melanoma Institute Australia (MIA) developed predictive models, but the utility of these models remains to be tested. To determine the clinical utility of the MIA and MSKCC models. This was a population-based comparative effectiveness research study including 10 089 consecutive patients with cutaneous melanoma undergoing SLNB from the Swedish Melanoma Registry from January 2007 to December 2021. Data were analyzed from May to August 2023. The predicted probability of SLN positivity was calculated using the MSKCC model and a limited MIA model (using mitotic rate as absent/present instead of count/mm2 and excluding the optional variable lymphovascular invasion) for each patient. The operating characteristics of the models were assessed and compared. The clinical utility of each model was assessed using decision curve analysis and compared with a strategy of performing SLNB on all patients. Among 10 089 included patients, the median (IQR) age was 64.0 (52.0-73.0) years, and 5340 (52.9%) were male. The median Breslow thickness was 1.8 mm, and 1802 patients (17.9%) had a positive SLN. Both models were well calibrated across the full range of predicted probabilities and had similar external area under the receiver operating characteristic curves (AUC; MSKCC: 70.8%; 95% CI, 69.5-72.1 and limited MIA: 69.7%; 95% CI, 68.4-71.1). At a risk threshold of 5%, decision curve analysis indicated no added net benefit for either model compared to performing SLNB for all patients. At risk thresholds of 10% or higher, both models added net benefit compared to SLNB for all patients. The greatest benefit was observed in patients with T2 melanomas using a threshold of 10%; in that setting, the use of the nomograms led to a net reduction of 8 avoidable SLNBs per 100 patients for the MSKCC nomogram and 7 per 100 patients for the limited MIA nomogram compared to a strategy of SLNB for all. This study confirmed the statistical performance of both the MSKCC and limited MIA models in a large, nationally representative data set. However, decision curve analysis demonstrated that using the models only improved selection for SLNB compared to biopsy in all patients when a risk threshold of at least 7% was used, with the greatest benefit seen for T2 melanomas at a threshold of 10%. Care should be taken when using these nomograms to guide selection for SLNB at the lowest thresholds.

Population-based validation of the melanoma institute australia (MIA) and the memorial sloan-kettering cancer center (MSKCC) predictive tool for sentinel node status in patients with melanoma.

9571 Background: Patients with primary cutaneous melanoma are selected for sentinel lymph node biopsy (SLNB) based upon their risk of a positive SLN being identified. To improve upon stage-based categorization, the Memorial Sloan Kettering Cancer Center (MSKCC) and Melanoma Institute Australia (MIA) developed predictive models. Which model is more clinically useful is currently unclear. Methods: Consecutive patients undergoing SLNB from 2007-2021 were identified from the Swedish Melanoma Registry (SweMR), which covers 99% of patients with invasive cutaneous melanoma in Sweden. The predicted probability of SLN positivity was calculated using the MSKCC and limited MIA models (using mitotic rate as absent/present instead of count and excluding the optional variable lymphovascular invasion) for each patient. The operating characteristics of the models were assessed and compared. The “clinical utility” of each model was assessed using decision curve analysis (DCA) and compared with a strategy of performing SLNB on all patients. Results: In total 10,089 patients were included; the median Breslow thickness was 1.8 mm, 33.7% were ulcerated, and 1,802 patients had a positive SLN (17.9%). Both the MSKCC and limited MIA models were well calibrated across the full range of predicted probabilities, and had similar external area under curve (AUC), MSKCC 70.8% (95% CI 69.5-72.1%) and MIA 70.3% (95% CI 68.9-71.7%). When new models were constructed to compare the incremental value of the additional parameters of either nomogram over Breslow thickness alone, the AUCs were 70.1%, 71.6% and 72.0% for Breslow alone, MIA parameters and MSKCC parameters, respectively. At a risk threshold of 5%, DCA indicated no added net benefit for either model compared to performing SLNB for all patients. At risk thresholds of 10% or higher, both models added net benefit compared to SLNB for all. The greatest benefit was observed from use of the models in patients with T2 melanomas when a threshold of 10% was used to select patients for SLNB. In that setting, use of the nomograms led to a net reduction in seven avoidable SLNBs per 100 patients for the limited MIA nomogram and eight for the MSKCC nomogram, when compared to a strategy of SLNB for all. Conclusions: This study confirms the statistical performance of both the MSKCC and limited MIA models in a large, nationally-representative dataset. However, clinical utility as assessed by DCA demonstrated that using the models only improved selection for SLNB when a risk threshold of at least 10% was employed, and then mainly for T2 melanomas. Either further development of the nomograms, or other strategies, are needed to improve selection at lower risk thresholds.

A New Model for Predicting Nonsentinel Lymph Node Metastasis in Early-Stage Breast Cancer Using MMP15.

Background In early-stage breast cancer (BC) patients, 40–70% of lymph node metastases are limited to the sentinel lymph nodes (SLNs). Patients at low risk for nonsentinel lymph node (NSLN) metastasis should be exempt from axillary lymph node dissection (ALND) or regional lymph node radiotherapy (RNI). Methods The present study included 237 female early-stage BC patients with positive SLNs who received ALND. Based on the clinicopathological factors of the 158 patients in the training cohort, multivariate analysis was used to determine the independent risk factors for NSLN metastasis, which were used to establish the NSLN metastasis prediction model. The calibration and discrimination of this model were tested with the training and validation cohorts and compared to the Memorial Sloan Kettering Cancer Center (MSKCC) model. Results Tumor size, neural invasion, lymphovascular invasion, expression of matrix metalloproteinase 15 (MMP15) in the cytoplasm, and the number of positive SLNs were statistically significant by multivariate analysis (P < 0.05), which were used to establish the new model. The MSKCC model was verified by the training cohort, and the area under the receiver-operating characteristic (ROC) curve was 0.733 (95% CI: 0.650–0.816), which was less than that of the new model (0.824; 95% CI: 0.760–0.889). The area under the ROC curve in the validation cohort for the new model was 0.773 (95% CI: 0.669–0.877), and the calibration performed well. The false-negative rates were 3.2%, 6.5%, and 14.5% for the predicted probability cut-offs of 50%, 60%, and 70%, respectively. Conclusions The new model included five variables: tumor size, neural invasion, lymphovascular invasion, cytoplasmic MMP15 expression, and the number of positive SLNs. The model with a cut-off of 60% could accurately identify low-risk patients with NSLN metastasis.

Machine Learning-Based Epigenetic Classifiers for Axillary Staging of Patients with ER-Positive Early-Stage Breast Cancer.

In the era of molecular stratification and effective multimodality therapies, surgical staging of the axilla is becoming less relevant for patients with estrogen receptor (ER)-positive early-stage breast cancer (EBC). Therefore, a nonsurgical method for accurately predicting lymph node disease is the next step in the de-escalation of axillary surgery. This study sought to identify epigenetic signatures in the primary tumor that accurately predict lymph node status. We selected a cohort of patients in The Cancer Genome Atlas (TCGA) with ER-positive, HER2-negative invasive ductal carcinomas, and clinically-negative axillae (n = 127). Clinicopathological nomograms from the Memorial Sloan Kettering Cancer Center (MSKCC) and the MD Anderson Cancer Center (MDACC) were calculated. DNA methylation (DNAm) patterns from primary tumor specimens were compared between patients with pN0 and those with > pN0. The cohort was divided into training (n = 85) and validation (n = 42) sets. Random forest was employed to obtain the combinations of DNAm features with the highest accuracy for stratifying patients with > pN0. The most efficient combinations were selected according to the area under the curve (AUC). Clinicopathological models displayed a modest predictive potential for identifying > pN0 disease (MSKCC AUC 0.76, MDACC AUC 0.69, p = 0.15). Differentially methylated sites (DMS) between patients with pN0 and those with > pN0were identified (n = 1656). DMS showed a similar performance to the MSKCC model (AUC = 0.76, p = 0.83). Machine learning approaches generated five epigenetic classifiers, which showed higher discriminative potential than the clinicopathological variables tested (AUC > 0.88, p < 0.05). Epigenetic classifiers based on primary tumor characteristics can efficiently stratify patients with no lymph node involvement from those with axillary lymph node disease, thereby providing an accurate method of staging the axilla.

Prognostic factors for patients with advanced renal cell carcinoma (aRCC) in the era of first-line (1L) treatment with immune checkpoint inhibitors (ICIs).

4544 Background: The most commonly used prognostic models in aRCC, the Memorial Sloan-Kettering Cancer Center (MSKCC) and the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) risk scores, were developed when cytokines and VEGF targeted monotherapies were standard of care, respectively. aRCC 1L treatment now includes combination therapy with 2 ICIs or ICI+VEGF receptor inhibitor. Re-evaluation of the MSKCC and IMDC prognostic models in the era of ICI therapy and identification of additional prognostic factors are overdue. Methods: Data from 1052 patients with aRCC treated on the CheckMate-214 (CM214) clinical trial with 1L nivolumab/ipilimumab (NIVO/IPI) or sunitinib (SUN) were analyzed retrospectively at median follow-up 5 yrs. For each treatment group, multivariable Cox model hazard ratios (HR) were compared to the published MSKCC and IMDC model HRs. Discrimination (c index) was assessed based upon the continuous scores obtained as the weighted combination of regression parameters from the published MSKCC &amp; IMDC models. Results: KPS &lt; 80% remained highly prognostic in both treatment groups (HRs &gt; 2.5). With the exception of high calcium, the other factors of time from dx to treatment &lt; 1 yr, hemoglobin &lt; lower limit of normal (LLN), neutrophils &gt; upper limit of normal (ULN), platelets &gt; ULN and LDH &gt; 1.5xULN consistently retained prognostic value for SUN (HRs ≥ published model HRs) but had diminished prognostic value for NIVO/IPI (each HR &lt; published model HR). High calcium had diminished prognostic value for SUN and retained prognostic value for NIVO/IPI. The c-indices for discrimination were 0.61 and 0.64 for the MSKCC model in the NIVO/IPI and SUN treatment groups, respectively, and were 0.63 and 0.67 for the IMDC model in the NIVO/IPI and SUN treatment groups compared to reported IMDC c-index of 0.73 (Heng 2009). (Table) Conclusions: The prognostic ability of the MSKCC and IMDC risk models is reduced with combination ICI treatment. Evaluation of additional prognostic factors and development of new risk scores are needed and this work is ongoing. [Table: see text]

Prognostic factors for overall survival in patients with clear cell metastatic renal cell carcinoma: Model development and external validation with Memorial Sloan Kettering Cancer Center model and the international metastatic renal cell carcinoma database consortium model.

To develop a new prognostic model for the overall survival of patients with clear cell metastatic renal cell carcinoma (mRCC) using Korean Renal Cancer Study Group (KRoCS) database and compared it with 2 renowned prognostic models: the Memorial Sloan Kettering Cancer Center (MSKCC) and the international metastatic renal cell carcinoma database consortium (IMDC) models.Data of 790 patients diagnosed with mRCC and receiving targeted therapy as their first-line treatment were pooled to this study. Data from 4 hospitals (n = 619) were used to develop the new model and those from other 5 hospitals (n = 171) were used for external validation. After detecting prognostic factors in multivariable Cox proportional-hazards regression analysis, patients were classified into 3 risk groups, favorable (0), intermediate (1–2), and poor (3 and more) by the number of prognostic factors.Seven variables such as more than 2 metastasis sites, no prior nephrectomy, Eastern Cooperative Oncology Group performance status ≥2, low hemoglobin, high serum corrected calcium, high neutrophil, high serum alkaline phosphatase were identified as prognostic factors for poor overall survival. Also, risk groups were categorized into 3 groups; median overall survival was 61.1 months in favorable, 26.5 months in intermediate, and 6.8 months in poor group. KRoCS ranked the first in all 3 statistical parameters including akaike information criterion (AIC), concordance index and generalized R2 among other prognostic models.We developed the KRoCS model and validated it externally with demonstrating its superiority over MSKCC and IMDC models. The KRoCS model can provide useful information for counseling patients with clear cell mRCC regarding life-expectancy.

Prognosis of Japanese metastatic renal cell carcinoma patients in the targeted therapy era.

The aims of this study were to investigate prognosis and validate prognostic models [Memorial Sloan-Kettering Cancer Center (MSKCC), International Metastatic Renal Cell Carcinoma Data Consortium (IMDC), and Japanese metastatic renal cancer (JMRC) models] in the targeted therapy era in Japanese patients with metastatic renal cell carcinoma. We retrospectively analyzed 692 patients who were diagnosed with mRCC from January 2008 to August 2018 in the Michinoku Japan Urological Cancer Study Group database. Nivolumab as sequential therapy was widely used. Other immune checkpoint inhibitors were excluded from this study. The median overall survival (95% confident interval) in all, MSKCC favorable, intermediate, and poor risk patients was 41.0months (33.9-46.8), not reached (63.5 to not estimable), 46.8months (37.1-52.9), and 10.4months (8.9-14.4), respectively. The median overall survival (95% confident interval) in IMDC favorable, intermediate, and poor risk patients was not reached (61.6 to not estimable), 47.4months (41.4-56.5), and 11.5 (9.9-16.3), respectively. The c-index of the MSKCC, IMDC, and JMRC models calculated at mRCC diagnosis was 0.680, 0.689, and 0.700, respectively. No statistical differences were found in the c-index among the models. While the real-world overall survival in Japanese patients with mRCC in the targeted therapy era improved compared to that previously reported in the cytokine era, there was no clear difference in the survival of poor risk patients between these eras. There were no differences in the superiority among the models.

Real-world outcomes in patients with metastatic renal cell carcinoma according to risk factors: the STAR-TOR registry.

Aim: Examine outcomes in sunitinib-treated patients by International Metastatic RCC Database Consortium (IMDC) or Memorial Sloan-Kettering Cancer Center (MSKCC) risk factors. Patients &methods: Patients enrolled in STAR-TOR registry (n=327). End points included overall survival, progression-free survivaland objective response rate. Results: Overall survival was similar for IMDC 0 versus 1 (p=0.238)or 2 versus ≥3 (p=0.156), but different for MSKCC (0 vs 1, p=0.037; 2 vs ≥3, p=0.001). Progression-free survival was similar for IMDC 2 versus 3 (p=0.306), but different for MSKCC (p=0.009). Objective response rate was different for IMDC 1 (41.9%) and 2 (29.5%) and similar for MSKCC 1 (34.4%) and 2 (31.0%). Conclusion: Outcome data varied according to IMDC or MSKCC. MSKCC model accurately stratify patients into risk groups. Clinical trial registration: NCT00700258 (ClinicalTrials.gov).

Application of a novel machine learning framework for predicting non-metastatic prostate cancer-specific mortality in men using the Surveillance, Epidemiology, and End Results (SEER) database

Accurate prognostication is crucial in treatment decisions made for men diagnosed with non-metastatic prostate cancer. Current models rely on prespecified variables, which limits their performance. We aimed to investigate a novel machine learning approach to develop an improved prognostic model for predicting 10-year prostate cancer-specific mortality and compare its performance with existing validated models. We derived and tested a machine learning-based model using Survival Quilts, an algorithm that automatically selects and tunes ensembles of survival models using clinicopathological variables. Our study involved a US population-based cohort of 171 942 men diagnosed with non-metastatic prostate cancer between Jan 1, 2000, and Dec 31, 2016, from the prospectively maintained Surveillance, Epidemiology, and End Results (SEER) Program. The primary outcome was prediction of 10-year prostate cancer-specific mortality. Model discrimination was assessed using the concordance index (c-index), and calibration was assessed using Brier scores. The Survival Quilts model was compared with nine other prognostic models in clinical use, and decision curve analysis was done. 647 151 men with prostate cancer were enrolled into the SEER database, of whom 171 942 were included in this study. Discrimination improved with greater granularity, and multivariable models outperformed tier-based models. The Survival Quilts model showed good discrimination (c-index 0·829, 95% CI 0·820-0·838) for 10-year prostate cancer-specific mortality, which was similar to the top-ranked multivariable models: PREDICT Prostate (0·820, 0·811-0·829) and Memorial Sloan Kettering Cancer Center (MSKCC) nomogram (0·787, 0·776-0·798). All three multivariable models showed good calibration with low Brier scores (Survival Quilts 0·036, 95% CI 0·035-0·037; PREDICT Prostate 0·036, 0·035-0·037; MSKCC 0·037, 0·035-0·039). Of the tier-based systems, the Cancer of the Prostate Risk Assessment model (c-index 0·782, 95% CI 0·771-0·793) and Cambridge Prognostic Groups model (0·779, 0·767-0·791) showed higher discrimination for predicting 10-year prostate cancer-specific mortality. c-indices for models from the National Comprehensive Cancer Care Network, Genitourinary Radiation Oncologists of Canada, American Urological Association, European Association of Urology, and National Institute for Health and Care Excellence ranged from 0·711 (0·701-0·721) to 0·761 (0·750-0·772). Discrimination for the Survival Quilts model was maintained when stratified by age and ethnicity. Decision curve analysis showed an incremental net benefit from the Survival Quilts model compared with the MSKCC and PREDICT Prostate models currently used in practice. A novel machine learning-based approach produced a prognostic model, Survival Quilts, with discrimination for 10-year prostate cancer-specific mortality similar to the top-ranked prognostic models, using only standard clinicopathological variables. Future integration of additional data will likely improve model performance and accuracy for personalised prognostics. None.

Development of a Novel Risk Score to Select the Optimal Candidate for Cytoreductive Nephrectomy Among Patients with Metastatic Renal Cell Carcinoma. Results from a Multi-institutional Registry (REMARCC)

BackgroundSelection of patients for upfront cytoreductive nephrectomy (CN) in metastatic renal cell carcinoma (mRCC) has to be improved. ObjectiveTo evaluate a new scoring system for the prediction of overall mortality (OM) in mRCC patients undergoing CN. Design, setting, and participantsWe identified a total of 519 patients with synchronous mRCC undergoing CN between 2005 and 2019 from a multi-institutional registry (Registry for Metastatic RCC [REMARCC]). Outcome measurements and statistical analysisCox proportional hazard regression was used to test the main predictors of OM. Restricted mean survival time was estimated as a measure of the average overall survival time up to 36 mo of follow-up. The concordance index (C-index) was used to determine the model’s discrimination. Decision curve analyses were used to compare the net benefit from the REMARCC model with International mRCC Database Consortium (IMDC) or Memorial Sloan Kettering Cancer Center (MSKCC) risk scores. Results and limitationsThe median follow-up period was 18 mo (interquartile range: 5.9–39.7). Our models showed lower mortality rates in obese patients (p = 0.007). Higher OM rates were recorded in those with bone (p = 0.010), liver (p = 0.002), and lung metastases (p < 0.001). Those with poor performance status (<80%) and those with more than three metastases had also higher OM rates (p = 0.026 and 0.040, respectively). The C-index of the REMARCC model was higher than that of the MSKCC and IMDC models (66.4% vs 60.4% vs 60.3%). After stratification, 113 (22.0%) patients were classified to have a favorable (no risk factors), 202 (39.5%) an intermediate (one or two risk factors), and 197 (38.5%) a poor (more than two risk factors) prognosis. Moreover, 72 (17.2%) and 51 (13.9%) patients classified as having an intermediate and a poor prognosis according to MSKCC and IMDC categories, respectively, would be reclassified as having a good prognosis according to the REMARCC score. ConclusionsOur findings confirm the relevance of tumor and patient features for the risk stratification of mRCC patients and clinical decision-making regarding CN. Further prospective external validations are required for the scoring system proposed herein. Patient summaryCurrent stratification systems for selecting patients for kidney removal when metastatic disease is shown are controversial. We suggest a system that includes tumor and patient features besides the systems already in use, which are based on blood tests.

External Validation of a Multiparametric Magnetic Resonance Imaging–based Nomogram for the Prediction of Extracapsular Extension and Seminal Vesicle Invasion in Prostate Cancer Patients Undergoing Radical Prostatectomy

The nomogram reported by Gandaglia et al (The key combined value of multiparametric magnetic resonance imaging, and magnetic resonance imaging-targeted and concomitant systematic biopsies for the prediction of adverse pathological features in prostate cancer patients undergoing radical prostatectomy. Eur Urol 2020;77:733–41) predicting extracapsular extension (ECE) or seminal vesicle invasion (SVI) has been developed using multiparametric magnetic resonance imaging (MRI) parameters and MRI-targeted biopsy. We aimed to validate this nomogram externally by analyzing 566 patients harboring prostate cancer diagnosed on MRI-targeted biopsy followed by radical prostatectomy. At final pathology, 37% and 12% patients had ECE and SVI, respectively. Performance of the nomogram, in comparison with the Memorial Sloan Kettering Cancer Center (MSKCC) model and Partin tables, was evaluated using discrimination, calibration, and decision curve analysis. Regarding ECE prediction, the nomogram showed higher discrimination (71.8% vs 69.8%, p = 0.3 and 71.8% vs 61.3%, p < 0.001), and similar miscalibration and net benefit for probability threshold above 30% when compared with MSKCC model and Partin tables, respectively. Performance of the nomogram with regard to SVI was comparable in terms of discrimination (68.5% vs 70.4% vs 67.8%, p ≥ 0.6), presenting a slight overestimation on calibration plots and a net benefit for probability threshold above 7.5%. This is the first multicentric study that externally validates a nomogram predicting ECE and SVI in patients diagnosed with MRI-targeted biopsy. Its performance was less optimistic than expected, and implementation of MRI in this setting was not associated with a clear improvement in patient selection and clinical usefulness when compared with available models. We proposed an updated version of the nomogram predicting ECE using the recalibration method, which leads to an improvement in its performance and needs to be validated in another external set. Patient summaryWe validate a prediction tool based on multiparametric magnetic resonance imaging (MRI) parameters and MRI-targeted biopsy predicting extracapsular extension and seminal vesicle invasion at radical prostatectomy. An improvement of patient selection was not clearly demonstrated when compared with available models based on clinical parameters, and implementation of MRI in this setting still needs to be clarified.

Development of novel ACN (albumin, C-reactive protein and neutrophil-to-lymphocyte ratio) prognostication model for patients with metastatic renal cell carcinoma receiving first-line molecular-targeted therapy

PurposeThe objective of this study was to develop a novel prognostication model in patients with treatment-naïve metastatic renal cell carcinoma (mRCC). MethodsThis study included 325 consecutive mRCC patients receiving first-line molecular-targeted therapy at 4 institutions. Potential parameters associated with overall survival (OS) in these patients were investigated to develop a novel stratification model. ResultsMedian OS of the 325 patients was 38 months. A multivariable analysis of several factors identified independent predictors associated with unfavorable OS as follows: no previous nephrectomy, Karnofsky performance status <80%, albumin (Alb) ≤3.5 g/dl, C-reactive protein (CRP) >0.5 mg/dl and neutrophil-to-lymphocyte ratio (NLR) >3. Of these 5 independent OS predictors, 3 numeric factors were used to develop the ACN (Alb, CRP, and NLR) model by dividing patients into 3 groups according to the positive numbers of these 3 numeric risk factors. Median OS durations were 63, 37, and 11 months in the favorable (n = 105, 32.3%, without risk factors), intermediate (n = 88, 27.1%, with a single risk factor), and poor (n = 132, 40.6%, with multiple risk factors) risk groups, respectively. The ACN model as a prognostication tool was shown to be superior to the Memorial Sloan Kettering Cancer Center (MSKCC) and the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) models by both the concordance index and decision curve analysis. ConclusionsThe ACN model could stratify the prognostic risk of mRCC patients receiving first-line targeted therapy more accurately than the MSKCC and IMDC models.

Improved Risk Prediction Calculator for Sentinel Node Positivity in Patients With Melanoma: The Melanoma Institute Australia Nomogram.

For patients with primary cutaneous melanoma, the risk of sentinel node (SN) metastasis varies according to several clinicopathologic parameters. Patient selection for SN biopsy can be assisted by National Comprehensive Cancer Network (NCCN) and ASCO/Society of Surgical Oncology (SSO) guidelines and the Memorial Sloan Kettering Cancer Center (MSKCC) online nomogram. We sought to develop an improved online risk calculator using alternative clinicopathologic parameters to more accurately predict SN positivity. Data from 3,477 patients with melanoma who underwent SN biopsy at Melanoma Institute Australia (MIA) were analyzed. A new nomogram was developed by replacing body site and Clark level from the MSKCC model with mitotic rate, melanoma subtype, and lymphovascular invasion. The predictive performance of the new nomogram was externally validated using data from The University of Texas MD Anderson Cancer Center (n = 3,496). The MSKCC model receiver operating characteristic curve had a predictive accuracy of 67.7% (95% CI, 65.3% to 70.0%). The MIA model had a predictive accuracy of 73.9% (95% CI, 71.9% to 75.9%), a 9.2% increase in accuracy over the MSKCC model (P < .001). Among the 2,748 SN-negative patients, SN biopsy would not have been offered to 22.1%, 13.4%, and 12.4% based on the MIA model, the MSKCC model, and NCCN or ASCO/SSO criteria, respectively. External validation generated a C-statistic of 75.0% (95% CI, 73.2% to 76.7%). A robust nomogram was developed that more accurately estimates the risk of SN positivity in patients with melanoma than currently available methods. The model only requires the input of 6 widely available clinicopathologic parameters. Importantly, the number of patients undergoing unnecessary SN biopsy would be significantly reduced compared with use of the MSKCC nomogram or the NCCN or ASCO/SSO guidelines, without losing sensitivity. An online calculator is available at www.melanomarisk.org.au.

Outcomes According to MSKCC Risk Score with Focus on the Intermediate-Risk Group in Metastatic Renal Cell Carcinoma Patients Treated with First-Line Sunitinib: A Retrospective Analysis of 2390 Patients.

Background: The Memorial Sloan–Kettering Cancer Center (MSKCC) prognostic model has been widely used for the prediction of the outcome of metastatic renal cell carcinoma (mRCC) patients treated with systemic therapies, however, data from large studies are limited. This study aimed at the evaluation of the impact of the MSKCC score on the outcomes in mRCC patients treated with first-line sunitinib, with a focus on the intermediate-risk group. Methods: Clinical data from 2390 mRCC patients were analysed retrospectively. Progression-free survival (PFS), overall survival (OS), and objective response rate (ORR) were analysed according to the MSKCC risk score. Results: ORR, median PFS, and OS for patients with one risk factor were 26.7%, 10.1, and 28.2 months versus 18.7%, 6.2, and 16.2 months, respectively, for those with two risk factors (ORR: p = 0.001, PFS: p < 0.001, OS: p < 0.001). ORR, median PFS, and OS were 33.0%, 17.0, and 44.7 months versus 24.1%, 9.0, and 24.1 months versus 13.4%, 4.5, and 9.5 months in the favourable-, intermediate-, and poor-risk groups, respectively (ORR: p < 0.001, PFS: p < 0.001, OS: p < 0.001). Conclusions: The results of the present retrospective study demonstrate the suitability of the MSKCC model in mRCC patients treated with first-line sunitinib and suggest different outcomes between patients with one or two risk factors.

A new prognostic score for disease progression and mortality in patients with newly diagnosed primary CNS lymphoma

BackgroundAlthough various prognostic models for primary central nervous system lymphoma (PCNSL) have been developed, there is no consensus regarding the optimal prognostic index. We aimed to evaluate potential prognostic factors and construct a novel predictive model for PCNSL patients.MethodsWe enrolled newly diagnosed PCNSL patients between 2003 and 2015. The primary endpoint was progression‐free survival (PFS), and the secondary endpoint was overall survival (OS). The prognostic factors identified using multivariate Cox proportional hazards models were used to develop a predictive model. We subsequently validated the prognostic model in an independent cohort. We also evaluated the validity of the existing scores: the International Extranodal Lymphoma Study Group (IELSG), the Nottingham/Barcelona (NB), and the Memorial Sloan‐Kettering Cancer Center models (MSKCC).ResultsWe identified 101 patients with newly diagnosed PCNSL at our center. Multivariate analysis showed that age ≥80, deep brain lesions, and ECOG ≥2 were independent risk factors of PFS. Assigning one point for each factor, we constructed a novel prognostic model, the Taipei Score, with four distinct risk groups (0‐3 points). The performances of the Taipei Score in discriminating both PFS and OS in the training cohort were significant, and the score was validated in the external validation cohort. The IELSG, NB and MSKCC models had insufficient discriminative ability for either PFS or OS in both cohorts.ConclusionThe Taipei Score is a simple model that discriminates PFS and OS for PCNSL patients. The score may offer disease risk stratification and facilitate clinical decision‐making.

A Prospective Validation Cohort Study of a Prediction Model on Non-sentinel Lymph Node Involvement in Early Breast Cancer.

Early breast cancer with one or two sentinel lymph nodes (SLNs) may omit axillary lymph node dissection (ALND) if followed by radiotherapy. However, only less than one-third of the patients have positive non-SLNs and can truly benefit from radiotherapy. Before any regional treatment decision, the risk of non-SLN metastasis must be identified. The authors previously developed a predictive model for non-SLN involvement using CK19 mRNA and contrast-enhanced ultrasound (CEUS) score in a training set. They designed a further study to evaluate the predictive effect using the model prospectively in a validation set of one or two involved SLNs. This study identified early breast cancer patients at Zhejiang Cancer Hospital from July 2017 to June 2018. The CK19 mRNA tested by quantitative real-time polymerase chain reaction and CEUS scores were collected before surgery. Patients with one or two involved SLNs were enrolled and underwent ALND. The estimated percentage of non-SLN involvement was calculated by the authors' model formula and the Memorial Sloan Kettering Cancer Center (MSKCC) nomogram. The false-negative rates, predictive accuracy, and area under curve (AUC) were compared between two predictive models. The study enrolled 235 patients, and 35.36% (83/235) of them had non-SLN involvement. The authors' model had a false-negative rate of 6% and an accuracy of 94.9%. The AUC was 0.952 (95% confidence interval [CI] 0.922-0.982), which was significantly higher than that of the MSKCC model at all three cutoff value levels. The authors' model, using CK19 mRNA and the CEUS score, showed the potential predictive value of non-SLNs before surgery for early breast cancer patients. NCT02992067, NCT03280134.

A prospective validation cohort study of a prediction model on non-sentinel lymph node involvement in early breast cancer.

559 Background: Early breast cancer (cT1-2N0) with one or two sentinel lymph node (SLN) involved may avoid axillary lymph node dissection (ALND) if follow by radiotherapy supported by Z0011 and AMAROS trials. However, only less than one-third of those patients have positive non-sentinel lymph node (nSLN) and can truly benefit from radiotherapy or ALND in those two trials. It is necessary to identify the risk of nSLN metastasis before local treatment decision. We previously developed a predictive model for nSLN involvement using circulating CK19 mRNA level combined with contrast-enhanced ultrasound (CEUS) score (ASCO2017 poster 239, NCT02992067) in a training set. To evaluate the predict effect of this model, we designed a further study using the model prospectively in a validation set (NCT03280134). Methods: We identified early breast cancer cases in Zhejiang Cancer Hospital from July 2017 to June 2018. The level of circulating CK19 mRNA tested by qRT-PCR and CEUS scores were collected before surgery in each case. Patients with 1~2 SLN involved were enrolled and continued for ALND. The estimated percentage of nSLN-involved were calculated both by our model formula and the Memorial Sloan-Kettering Cancer Center (MSKCC) nomogram. The predictive accuracy and false negative rates (FNR) were evaluated and the area under curve (AUC) was compared between two predictive models. Results: Totally, 235 patients diagnosed as early breast cancer with 1~2 SLN involved were enrolled and 35.36% of them were nSLN involved after ALND. The total accuracy and FNR by our model for nSLN-involved prediction was 94.89% and 6.02%, respectively. The AUC was 0.952 (95%CI, 0.922~0.982), significantly higher than that in MSKCC model 0.880 (95%CI, 0.833~0.927). Furthermore, only CK19 mRNA level (HR = 40.091, 95%CI, 13.663~117.635) and CEUS score (HR = 2.009, 95%CI 1.158~3.485) are significantly related to nSLN involvement in both univariate and multivariate analysis, adjusted by age, menopause statue, tumor size, histological grade, estrogen receptor, progesterone receptor and human epidermalgrowth factor receptor-2 expression. Conclusions: Our model using CK19 mRNA and CEUS score showed potential predictive value of nSLN before surgery in early breast cancer patients. Further validation in larger multicenter cohort is warranted before changing clinical practice.

Impact of disagreement between the IMDC and MSKCC risk groups on prognosis in patients with metastatic renal cell carcinoma.

556 Background: As the clinical implication of the risk group disagreement between the risk models remains unclear, we aimed to investigate the impact of the risk group disagreement between the Memorial Sloan Kettering Cancer Center (MSKCC) and the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) models on prognosis. Methods: We retrospectively evaluated 176 patients with metastatic renal cell carcinoma (mRCC) who were treated with tyrosine kinase inhibitors as first-line therapy in five hospitals between October 2008 and August 2018. The risk group classification differences between the MSKCC and the IMDC models were evaluated using criteria of agreement (identical risk group in both the MSKCC and IMDC models) and disagreement (not identical risk group in both the MSKCC and IMDC models). The agreement of risk stratification between the MSKCC and IMDC models was evaluated using Cohen’s k coefficient. Oncological outcomes were compared between the agreement and disagreement groups. Results: The number of patients with agreement, upgrade, and downgrade was 135/176 (77%), 39/176 (22%), and 2/176 (1.1%), respectively. Of 41 patients with disagreement, reclassification from the MSKCC-intermediate to the IMDC-poor-risk group was most frequent (n = 34/176, 19%). The Cohen’s k coefficient for agreement of the two risk models was substantial with k value of 0.613 ( P &lt; 0.001). Significantly poorer prognosis was observed in patients with disagreement than in those with agreement. Conclusions: Disagreement between the MSKCC and IMDC models may have a negative impact on prognosis in patients with mRCC. Further study is necessary to validate our findings.

Impact of Disagreement Between Two Risk Group Models on Prognosis in Patients With Metastatic Renal-Cell Carcinoma.

To investigate the impact of the risk group disagreement between the Memorial Sloan Kettering Cancer Center (MSKCC) and the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) models on prognosis. We retrospectively evaluated 176 patients with metastatic renal-cell carcinoma who were treated with tyrosine kinase inhibitors as first-line therapy in 5 hospitals between October 2008 and August 2018. The risk group classification differences between the MSKCC and the IMDC models were evaluated using criteria of agreement (identical risk group in both the MSKCC and IMDC models) and disagreement (not identical risk group in both the MSKCC and IMDC models). The agreement of risk stratification between the models was evaluated using the Cohen κ coefficient. Oncologic outcomes were compared between the agreement and disagreement groups. The number of patients with agreement, upgrade, and downgrade was 135 (77%), 39 (22%), and 2 (1.1%), respectively. Of 41 patients with disagreement, reclassification from the MSKCC-intermediate to the IMDC-poor risk group was most frequent (n= 34, 19%). The Cohen κ coefficient for agreement was substantial, with κ= 0.613 (P< .001). Significantly poorer prognosis was observed in patients with disagreement than in those with agreement. Neutrophil count, hemoglobin, serum calcium concentration, and C-reactive protein were significantly different between the groups. Disagreement between the MSKCC and IMDC models may have a negative impact on prognosis in patients with metastatic renal-cell carcinoma. The inclusion of systematic inflammation markers in a risk model may be essential for prognosis prediction.

Genomically annotated risk model for advanced renal-cell carcinoma: a retrospective cohort study.

The Memorial Sloan Kettering Cancer Center (MSKCC) risk model is an established prognostic tool for metastatic renal-cell carcinoma that integrates clinical and laboratory data, but is agnostic to tumour genomics. Several mutations, including BAP1 and PBRM1, have prognostic value in renal-cell carcinoma. Using two independent clinical trial datasets of patients with metastatic renal-cell carcinoma, we aimed to study whether the addition of the mutation status for several candidate prognostic genes to the MSKCC model could improve the model's prognostic performance. In this retrospective cohort study, we used available formalin-fixed paraffin-embedded tumour tissue and clinical outcome data from patients with metastatic renal-cell carcinoma assigned to treatment with tyrosine kinase inhibitors in the COMPARZ trial (training cohort; n=357) and RECORD-3 trial (validation cohort; n=258). Eligible patients in both trials were treatment-naive; had histologically confirmed, advanced, or metastatic renal-cell carcinoma; and a Karnofsky performance status score of at least 70. For each cohort, data from patients in all treatment groups (sunitinib and pazopanib in the training cohort, and everolimus and sunitinib in the validation cohort) were pooled for this analysis. In the training cohort, tumour tissue was used to evaluate somatic mutations by next-generation sequencing, and the association between cancer-specific outcomes (overall survival, progression-free survival, and overall response) and the mutation status of six genes of interest (BAP1, PBRM1, TP53, TERT, KDM5C, and SETD2) was tested. Only those genes with prognostic value in this setting were added to the MSKCC risk model to create a genomically annotated version. The validation cohort was used to independently test the prognostic value of the annotated model compared with the original MSKCC risk model. 357 (32%) of 1110 patients assigned to protocol treatment in the COMPARZ study between August, 2008, and September, 2011, were evaluable for mutation status and clinical outcomes in the training cohort. The independent validation cohort included 258 (55%) of 471 evaluable patients, enrolled between October, 2009, and June, 2011, on the RECORD-3 study. In the training cohort, the presence of any mutation in BAP1 or TP53, or both, and absence of any mutation in PBRM1 were prognostic in terms of overall survival (TP53wt/BAP1mut, TP53mut/BAP1wt o TP53mut/BAP1mut vs TP53wt/BAP1wt hazard ratio [HR] 1·57, 95% CI 1·21-2·04; p=0·0008; PBRM1wt vs PBRMmut, HR 1·58, 1·16-2·14; p=0·0035). The mutation status for these three prognostic genes were added to the original MSKCC risk model to create a genomically annotated version. Distribution of participants in the training cohort into the three risk groups of the original MSKCC model changed from 87 (24%) of 357 patients deemed at favourable risk, 217 (61%) at intermediate risk, and 53 (15%) at poor risk, to distribution across four risk groups in the genomically annotated risk model, with 36 (10%) of 357 deemed at favourable risk, 77 (22%) at good risk, 108 (30%) at intermediate risk, and 136 (38%) at poor risk. Addition of genomic information improved model performance for predicting overall survival (C-index: original model, 0·595 [95% CI 0·557-0·634] vs new model, 0·637 [0·595-0·679]) and progression-free survival (0·567 [95% CI 0·529-0·604] vs 0·602 [0·560-0·643]) with adequate discrimination of the proportion of patients who achieved an objective response (Cochran-Armitage one-sided p=0·0014). Analyses in the validation cohort confirmed the superiority of the genomically annotated risk model over the original version. The mutation status of BAP1, PBRM1, and TP53 has independent prognostic value in patients with advanced or metastatic renal-cell carcinoma treated with first-line tyrosine kinase inhibitors. Improved stratification of patients across risk groups by use of a genomically annotated model including the mutational status of these three genes warrants further investigation in prospective trials and could be of use as a model to stratify patients with metastatic renal-cell carcinoma in clinical trials. Novartis Pharmaceuticals Corporation, MSKCC Support Grant/Core Grant, and the J Randall & Kathleen L MacDonald Research Fund.