Concordance Probability Estimate Research Articles

Comparison of the prognostic value of eight nutrition-related tools in older patients with cancer: A prospective study

ObjectivesThe primary objective of the present study was to evaluate and compare the ability of eight nutrition-related tools to predict 1-year mortality in older patients with cancer. Design, setting and participantsWe studied older patients with cancer from the ELCAPA cohort and who had been referred for a geriatric assessment at one of 14 participating geriatric oncology clinics in the greater Paris area of France between 2007 and 2018. MeasurementsThe studied nutrition-related tools/markers were the body mass index (BMI), weight loss (WL) in the previous 6 months, the Mini Nutritional Assessment, the Geriatric Nutritional Risk Index (GNRI), the Prognostic Nutritional Index, the Glasgow Prognostic Score (GPS), the modified GPS, and the C-reactive protein/albumin ratio. ResultsA total of 1361 patients (median age: 81; males: 51%; metastatic cancer: 49%) were included in the analysis. Most of the tools showed a progressively increase in the mortality risk as the nutrition-related risk category worsened (overall p-values <0.02 for all) after adjustment for age, outpatient status, functional status, severe comorbidities, cognition, mood, cancer treatment strategy, tumour site, and tumour metastasis. All the models were discriminant, with a C-index ranging from 0.748 (for the BMI) to 0.762 (for the GPS). The concordance probability estimate ranged from 0.764 (WL) to 0.773 (GNRI and GPS)). ConclusionAfter adjustment for relevant prognostic factors, all eight nutrition-related tools/markers were independently associated with 1-year mortality in older patients with cancer. Depending on the time or context of the GA, physicians do not always have the time or means to perform and assess all the tools/markers compared here. However, even when some information is missing, each nutritional tool/marker has prognostic value and can be used in the evaluation.

Accounting for time dependency in meta-analyses of concordance probability estimates.

Recent years have seen the development of many novel scoring tools for disease prognosis and prediction. To become accepted for use in clinical applications, these tools have to be validated on external data. In practice, validation is often hampered by logistical issues, resulting in multiple small-sized validation studies. It is therefore necessary to synthesize the results of these studies using techniques for meta-analysis. Here we consider strategies for meta analyzing the concordance probability for time-to-event data ("C-index"), which has become a popular tool to evaluate the discriminatory power of prediction models with a right-censored outcome. We show that standard meta-analysis of the C-index may lead to biased results, as the magnitude of the concordance probability depends on the length of the time interval used for evaluation (defined e.g., by the follow-up time, which might differ considerably between studies). To address this issue, we propose a set of methods for random-effects meta-regression that incorporate time directly as covariate in the model equation. In addition to analyzing nonlinear time trends via fractional polynomial, spline, and exponential decay models, we provide recommendations on suitable transformations of the C-index before meta-regression. Our results suggest that the C-index is best meta-analyzed using fractional polynomial meta-regression with logit-transformed C-index values. Classical random-effects meta-analysis (not considering time as covariate) is demonstrated to be a suitable alternative when follow-up times are small. Our findings have implications for the reporting of C-index values in future studies, which should include information on the length of the time interval underlying the calculations.

Prognostic Value of Tumor Volume Assessment on PSMA PET After 177Lu-PSMA Radioligand Therapy Evaluated by PSMA PET/CT Consensus Statement and RECIP 1.0.

Quantitative evaluation of prostate-specific membrane antigen (PSMA)-targeting PET/CT remains challenging but is urgently needed for the use of standardized PET-based response criteria, such as the PSMA PET/CT consensus statement or Response Evaluation Criteria in PSMA PET/CT (RECIP 1.0). A recent study evaluated the prognostic value of whole-body tumor volume using a semiautomatic method relying on a 50% threshold of lesion SUVmax (PSMATV50). In the present study, we analyzed the suitability of this approach comparing 18F-PSMA-1007 with 68Ga-PSMA-11 PET/CT scans and the potential of PSMATV50 for the prediction of overall survival (OS) in patients before 177Lu-PSMA radioligand therapy (RLT). Moreover, PSMATV50 was integrated into the PSMA PET/CT consensus statement as well as RECIP 1.0, and the prognostic value of these response classification systems was compared. Methods: This retrospective study included 70 patients with metastatic castration-resistant prostate cancer undergoing PSMA RLT. Thirty-three patients were monitored by 68Ga-PSMA-11 PET/CT, and 37 patients by 18F-PSMA-1007 PET/CT. PET/CT scans before (baseline) and at the end of PSMA RLT after 2-4 cycles (follow-up) were separately analyzed by 2 readers. PSMATV50 at baseline and its change at the time of follow-up (ΔPSMATV50, expressed as a ratio) were correlated with OS using Cox proportional-hazards regression. The results of both subgroups were compared. The integration of ΔPSMATV50 in existing response classification systems was evaluated. To assess and compare the discriminatory strength of these classification systems, Gönen and Heller concordance probability estimates were calculated. Results: PSMATV50 determination was technically feasible in all examinations. A higher PSMATV50 at baseline and a higher ΔPSMATV50 were strongly associated with a shorter OS for both 68Ga-PSMA-11 (PSMATV50: hazard ratio [HR] of 1.29 [95% CI, 1.05-1.55], P = 0.009; ΔPSMATV50: HR of 1.83 [95% CI, 1.08-3.09], P = 0.024) and 18F-PSMA-1007 (PSMATV50: HR of 1.84 [95% CI, 1.13-2.99], P = 0.014; ΔPSMATV50: HR of 1.23 [95% CI, 1.04-1.51], P = 0.03). Response assessment provided high discriminatory power for OS for the PSMA PET/CT consensus statement (concordance probability estimate, 0.73) as well as RECIP 1.0 (concordance probability estimate, 0.74). Conclusion: PSMATV50 and ΔPSMATV50 proved to be predictive of OS not only for 68Ga-PSMA-11 but also for 18F-PSMA-1007 PET/CT scans. Subsequent integration of ΔPSMATV50 into the PSMA PET/CT consensus statement and RECIP 1.0 provided equally high prognostic value for both classification systems.

Tumor-Infiltrating Neutrophils after Neoadjuvant Therapy are Associated with Poor Prognosis in Esophageal Cancer

BackgroundIn esophageal cancer (EC), there is a paucity of knowledge regarding the interplay between the tumor immune microenvironment and response to neoadjuvant treatment and, therefore, which factors may influence outcomes. Thus, our goal was to investigate the changes in the immune microenvironment with neoadjuvant treatment in EC by assessing the expression of immune related genes and their association with prognosis.MethodsWe examined the transcriptome of paired pre- and post-neoadjuvant treated EC specimens. Based on these findings, we validated the presence of tumor-infiltrating neutrophils using CD15+ immunohistochemistry in a discovery cohort of patients with residual pathologic disease. We developed a nomogram as a predictor of progression-free survival (PFS) incorporating the variables CD15+ cell count, tumor regression grade, and tumor grade.ResultsAfter neoadjuvant treatment, there was an increase in genes related to myeloid cell differentiation and a poor prognosis associated with high neutrophil (CD15+) counts. Our nomogram incorporating CD15+ cell count was predictive of PFS with a C-index of 0.80 (95% confidence interval [CI] 0.68–0.9) and a concordance probability estimate (CPE) of 0.77 (95% CI 0.69–0.86), which indicates high prognostic ability. The C-index and CPE of the validation cohort were 0.81 (95% CI 0.69–0.91) and 0.78 (95% CI 0.7–0.86), respectively.ConclusionsOur nomogram incorporating CD15+ cell count can potentially be used to identify patients at high risk of recurrent disease and thus stratify patients who will benefit most from adjuvant treatment.

Small cell neuroendocrine carcinoma of the cervix: Analysis of prognostic factors and patterns of metastasis

ObjectivesTo describe characteristics and outcomes of patients with small cell neuroendocrine carcinoma of the cervix (SCNCC) and determine the staging system most predictive of outcome—the two-tier (limited-stage [LS] vs. extensive-stage [ES]) or International Federation of Gynecology and Obstetrics (FIGO) staging system. MethodsPatients with SCNCC evaluated at our institution from 1/1/1990–6/30/2021 were included. Medical records were reviewed for variables of interest. Appropriate statistical tests were performed to determine associations. Survival curves were created using the Kaplan-Meier method. Concordance probability estimates (CPEs) were calculated to evaluate the prediction probability of the staging systems. ResultsOf 63 patients, 41 had LS and 22 ES SCNCC. Patients with ES disease were significantly older than those with LS disease (median, 54 and 37 years, respectively; p < 0.001). Smoking status, race, and history of HPV were not associated with stage or outcomes. Forty-eight patients had metastatic disease (24 [50%] at initial diagnosis). The most common first sites of metastasis were lung (n = 20/48, 42%), lymph nodes (n = 19/48, 40%), and liver (n = 13/48, 27%). Nine patients had brain metastasis (8 symptomatic at recurrence; 1 asymptomatic at initial diagnosis). Both staging systems were associated with progression-free and overall survival. Adjusted CPE found the FIGO staging system was more predictive of outcomes than the two-tier staging system. ConclusionsProviders should have a low threshold to obtain brain imaging for patients with SCNCC, especially in the presence of visceral metastases. FIGO staging should be used to classify SCNCC. Further research is necessary to understand prognostic factors of this rare disease.

Evaluation of the prognostic value of the ESTRO EORTC classification of oligometastatic disease in patients treated with stereotactic body radiotherapy: A retrospective single center study

PurposeTo explore the prognostic value of the oligometastatic disease (OMD) states as proposed by the European Society for Radiotherapy and Oncology (ESTRO) European Organisation for Research and Treatment of Cancer (EORTC) classification system. Materials and MethodsThis retrospective single-institution study included patients with 1–5 extracranial metastases from any solid malignancy treated with SBRT to all metastases. OMD states were defined according to the ESTRO EORTC classification. Overall survival (OS) and progression-free survival (PFS) were analyzed using the Kaplan–Meier method. Discriminatory strength of the classification was assessed by Gönen & Heller’s concordance probability estimate (CPE). Univariable and multivariable Cox regression models were used to assess predictors of OS and PFS. ResultsIn total, 385 patients were included. The median follow-up was 24.1 months. The most frequent OMD states were metachronous oligorecurrence (23.6%) and induced oligoprogression (18.7%). Induced OMD patients had significantly shorter median OS (28.1 months) compared with de-novo (46.3 months, p = 0.002) and repeat OMD (50.3 months, p = 0.002). Median PFS in de-novo OMD patients (8.8 months) was significantly longer than in repeat (5.4 months, p = 0.002) and induced OMD patients (4.3 months, p < 0.001). The classification system had moderate discriminatory strength for OS and PFS. Multivariable analyses confirmed that compared with induced OMD, de-novo OMD was associated with longer PFS and repeat with longer OS. ConclusionAll patients were successfully categorized according to the ESTRO EORTC classification system. The discriminatory strength of the classification was confirmed for OMD patients treated with metastases-directed SBRT. Larger multicenter trials are needed to validate the prognostic power for OMD patients irrespective of primary tumor and treatment approach.

Assessing the accuracy of the lung allocation score

The United States (US) Lung Allocation Score (LAS) relies on the performance of 2 survival models that estimate waitlist and post-transplant survival. These models were developed using data from 2005 to 2008, and it is unknown if they remain accurate. We performed an observational cohort study of US lung transplantation candidates and recipients greater than 12 years of age between February 19, 2015 and February 19, 2019. We evaluated the LAS waitlist and post-transplant models with the concordance probability estimate and by comparing predicted vs observed 1-year restricted mean survival times by risk decile. We then compared a nonparametric estimate of the observed LAS with the predicted LAS for each percentile of recipients. The waitlist model ranked candidates (N=11,539) in the correct risk order 72% of the time (95% CI 71%-73%), and underestimated candidate one-year survival by 136 days for the highest risk decile (p < 0.001). The post-transplant model ranked recipients (N=9,377) in the correct risk order 57% of the time (95% CI 55-58%), and underestimated recipient one-year survival by 70 days for the highest risk decile (p < 0.001). Overall, the LAS at transplant explained only 56% of the variation in observed outcomes, and was increasingly inaccurate at higher predicted values. The waitlist and the post-transplant models that constitute the LAS are inaccurate, limiting the ability of the system to rank candidates on the waitlist in the correct order. The LAS should therefore be updated and the underlying models should be modernized.

DXA Versus Clinical Measures of Adiposity as Predictors of Cardiometabolic Diseases and All-Cause Mortality in Postmenopausal Women

ObjectiveTo investigate whether dual-energy x-ray absorptiometry (DXA) estimates of adiposity improve risk prediction for cardiometabolic diseases over traditional surrogates, body mass index (BMI), waist circumference (WC), and waist-to-hip ratio (WHR) in older women. Patients and MethodsWe analyzed up to 9744 postmenopausal women aged 50 to 79 years participating in the Women’s Health Initiative who underwent a DXA scan and were free of cardiovascular disease and diabetes at baseline (October 1993 to December 1998) and followed through September 2015. Baseline BMI, WC, WHR, and DXA-derived percent total-body and trunk fat (%TrF) were incorporated into multivariable Cox proportional hazards models to estimate the risk of incident diabetes, atherosclerosis-related cardiovascular diseases (ASCVDs), heart failure, and death. Concordance probability estimates assessed the relative discriminatory value between pairs of adiposity measures. ResultsA total of 1327 diabetes cases, 1266 atherosclerotic cardiovascular disease (ASCVD) cases, 292 heart failure cases, and 1811 deaths from any cause accrued during a median follow-up of up to 17.2 years. The largest hazard ratio observed per 1 standard deviation increase of an adiposity measure was for %TrF and diabetes (1.77; 95% CI, 1.66-1.88) followed by %TrF and broadly defined ASCVD (1.22; 95% CI, 1.15-1.30). These hazard ratios remained significant for both diabetes (1.47; 95% CI, 1.37-1.57) and ASCVD (1.22; 95% CI, 1.14-1.31) even after adjusting for the best traditional surrogate measure of adiposity, WC. Percentage of trunk fat was also the only adiposity measure to demonstrate statistically significant improved concordance probability estimates over BMI, WC, and WHR for diabetes and ASCVD (all P<0.05). ConclusionDXA-derived estimates of abdominal adiposity in postmenopausal women may allow for substantially improved risk prediction of diabetes over standard clinical risk models. Larger DXA studies with complete lipid biomarker profiles and clinical trials are needed before firm conclusions can be made.

Development and Validation of a Novel Nomogram for Individualized Prediction of Survival in Cancer of Unknown Primary.

Prognostic uncertainty is a major challenge for cancer of unknown primary (CUP). Current models limit a meaningful patient-provider dialogue. We aimed to establish a nomogram for predicting overall survival (OS) in CUP based on robust clinicopathologic prognostic factors. We evaluated 521 patients with CUP at MD Anderson Cancer Center (MDACC; Houston, TX; 2012-2016). Baseline variables were analyzed using Cox regression and nomogram developed using significant predictors. Predictive accuracy and discriminatory performance were assessed by calibration curves, concordance probability estimate (CPE ± SE), and concordance statistic (C-index). The model was subjected to bootstrapping and multi-institutional external validations using two independent CUP cohorts: V1 [MDACC (2017), N = 103] and V2 (BC Cancer, Vancouver, Canada and Sarah Cannon Cancer Center/Tennessee Oncology, Nashville, TN; N = 302). Baseline characteristics of entire cohort (N = 926) included: median age (63 years), women (51%), Eastern Cooperative Oncology Group performance status (ECOG PS) 0-1 (64%), adenocarcinomas (52%), ≥3 sites of metastases (30%), and median follow-up duration and OS of 40.1 and 14.7 months, respectively. Five independent prognostic factors were identified: gender, ECOG PS, histology, number of metastatic sites, and neutrophil-lymphocyte ratio. The resulting model predicted OS with CPE of 0.69 [SE: ± 0.01; C-index: 0.71 (95% confidence interval: 0.68-0.74)] outperforming Culine/Seve prognostic models (CPE: 0.59 ± 0.01). CPE for external validation cohorts V1 and V2 were 0.67 (± 0.02) and 0.70 (± 0.01), respectively. Calibration curves for 1-year OS showed strong agreement between nomogram prediction and actual observations in all cohorts. Our user-friendly CUP nomogram integrating commonly available baseline factors provides robust personalized prognostication which can aid clinical decision making and selection/stratification for clinical trials.

Assessing the Accuracy of the Lung Allocation Score

<h3>Purpose</h3> The Lung Allocation Score (LAS) relies on the performance of two cox proportional hazards models that estimate waitlist and post-transplant survival. These models were developed using data from 2005-2008; it is unknown if they remain accurate. <h3>Methods</h3> Observational cohort study of all lung transplant candidates and recipients greater than 12 years listed or transplanted between February, 19^th 2015 to February, 19^th 2019 in the Scientific Registry of Transplant Recipients database. We evaluated the discrimination of the waitlist and post-transplant models with the concordance probability estimate and the calibration of each model by comparing predicted versus observed one-year restricted mean survival times. We evaluated the overall accuracy of LAS by comparing a non-parametric estimate of the observed LAS with the actual LAS at transplant for each recipient percentile. <h3>Results</h3> During the study period, 11,539 eligible candidates were listed for transplant and 9,377 eligible recipients were transplanted. The waitlist model ranked candidates in the correct risk order 72% of the time (95% CI 71 - 73%) and underestimated survival by 59 days (95% CI 57 - 61 days) on average. The miscalibration increased from 6 days of error (95% CI 4 - 9 days) in the lowest risk decile to 136 days (95% CI 107-164 days) in the highest risk decile. The post-transplant model ranked recipients in the correct risk order 57% of the time (95% CI 55 - 58%) and underestimated post-transplant survival by 25 days (95% CI 23-27 days) on average. The miscalibration increased from 10 days of error in the lowest risk decile (95% CI 7 - 14 days) to 70 days (95% CI 64 - 77) in the highest risk decile. Overall, a recipient's LAS at transplant explained only 56% of variation in the observed LAS. <h3>Conclusion</h3> The waitlist and the post-transplant models that form the backbone of LAS are inaccurate, limiting the ability of the system to rank candidates in the correct order. The LAS should be updated using a modern cohort patient cohort.

Improved survival prediction for oropharyngeal cancer beyond TNMv8

PurposeFor oropharynx squamous cell carcinoma (OPSCC) this study aimed to: (i) compare 5-year overall survival (OS) stratification by AJCC/UICC TNM versions 7 (TNMv7) and 8 (TNMv8), (ii) determine whether changes to T and N stage groupings improve prognostication and (iii) develop and validate a model incorporating additional clinical characteristics to improve 5-year OS prediction. Material and MethodsAll OPSCC treated with curative-intent at our institution between 2011 and 2017 were included. The primary endpoint was 5-year OS. Survival curves were produced for TNMv7 and TNMv8. A three-way interaction between T, N stage and p16 status was evaluated for improved prognostication. Cox proportional hazards modelling was used to derive a new predictive model. ResultsOf 750 OPSCC cases, 574 (77%) were p16-positive. TNMv8 was more prognostic than TNMv7 (concordance probability estimate [CPE] ± SE = 0.72 ± 0.02 vs 0.53 ± 0.02). For p16-positive disease, TNMv8 discriminated stages II vs I (HR 2.32, 95% CI 1.47–3.67) and III vs II (HR 1.75, 95% CI 1.13–2.72). For p16-negative disease, TNMv7 and TNMv8 demonstrated poor hazard discrimination. Different T, N stage and p16-status combinations did not improve prognostication after adjusting for other factors (CPE = 0.79 vs 0.79, p = 0.998). A model for p16-positive and p16-negative OPSCC including additional clinical characteristics improved 5-year OS prediction beyond TNMv8 (c-index 0.76 ± 0.02). ConclusionsTNMv8 is superior to TNMv7 for p16-positive OPSCC, but both performed poorly for p16-negative disease. A novel model incorporating additional clinical characteristics improved 5-year OS prediction for both p16-positive and p16-negative disease.

The optimal cut-off values for tumor size, number of lesions, and CEA levels in patients with surgically treated colorectal cancer liver metastases: An international, multi-institutional study.

Despite the long-standing consensus on the importance of tumor size, tumor number and carcinoembryonic antigen(CEA) levels as predictors of long-term outcomes among patients with colorectal liver metastases (CRLM), optimal prognostic cut-offs for these variables have not been established. Patients who underwent curative-intent resection of CRLM and had available data on at least one of the three variables of interest above were selected from a multi-institutional dataset of patients with known KRAS mutational status. The resulting cohort was randomly split into training and testing datasets and recursive partitioning analysis was employed to determine optimal cut-offs. The concordance probability estimates (CPEs) for these optimal cut offs were calculated and compared to CPEs for the most widely used cut-offs in the surgical literature. A total of 1643 patients who met eligibility criteria were identified. Following recursive partitioning analysis in the training dataset, the following cut-offs were identified: 2.95 cm for tumor size, 1.5 for tumor number and 6.15 ng/ml for CEA levels. In the entire dataset, the calculated CPEs for the new tumor size (0.52), tumor number (0.56) and CEA (0.53) cut offs exceeded CPEs for other commonly employed cut-offs. The current study was able to identify optimal cut-offs for the three most commonly employed prognostic factors in CRLM. While the per variable gains in discriminatory power are modest, these novel cut-offs may help produce appreciable increases in prognostic performance when combined in the context of future risk scores.

A Genomic-Pathologic Annotated Risk Model to Predict Recurrence in Early-Stage Lung Adenocarcinoma

Recommendations for adjuvant therapy after surgical resection of lung adenocarcinoma (LUAD) are based solely on TNM classification but are agnostic to genomic and high-risk clinicopathologic factors. Creation of a prediction model that integrates tumor genomic and clinicopathologic factors may better identify patients at risk for recurrence. To identify tumor genomic factors independently associated with recurrence, even in the presence of aggressive, high-risk clinicopathologic variables, in patients with completely resected stages I to III LUAD, and to develop a computational machine-learning prediction model (PRecur) to determine whether the integration of genomic and clinicopathologic features could better predict risk of recurrence, compared with the TNM system. This prospective cohort study included 426 patients treated from January 1, 2008, to December 31, 2017, at a single large cancer center and selected in consecutive samples. Eligibility criteria included complete surgical resection of stages I to III LUAD, broad-panel next-generation sequencing data with matched clinicopathologic data, and no neoadjuvant therapy. External validation of the PRecur prediction model was performed using The Cancer Genome Atlas (TCGA). Data were analyzed from 2014 to 2018. The study end point consisted of relapse-free survival (RFS), estimated using the Kaplan-Meier approach. Associations among clinicopathologic factors, genomic alterations, and RFS were established using Cox proportional hazards regression. The PRecur prediction model integrated genomic and clinicopathologic factors using gradient-boosting survival regression for risk group generation and prediction of RFS. A concordance probability estimate (CPE) was used to assess the predictive ability of the PRecur model. Of the 426 patients included in the analysis (286 women [67%]; median age at surgery, 69 [interquartile range, 62-75] years), 318 (75%) had stage I cancer. Association analysis showed that alterations in SMARCA4 (clinicopathologic-adjusted hazard ratio [HR], 2.44; 95% CI, 1.03-5.77; P = .042) and TP53 (clinicopathologic-adjusted HR, 1.73; 95% CI, 1.09-2.73; P = .02) and the fraction of genome altered (clinicopathologic-adjusted HR, 1.03; 95% CI, 1.10-1.04; P = .005) were independently associated with RFS. The PRecur prediction model outperformed the TNM-based model (CPE, 0.73 vs 0.61; difference, 0.12 [95% CI, 0.05-0.19]; P < .001) for prediction of RFS. To validate the prediction model, PRecur was applied to the TCGA LUAD data set (n = 360), and a clear separation of risk groups was noted (log-rank statistic, 7.5; P = .02), confirming external validation. The findings suggest that integration of tumor genomics and clinicopathologic features improves risk stratification and prediction of recurrence after surgical resection of early-stage LUAD. Improved identification of patients at risk for recurrence could enrich and enhance accrual to adjuvant therapy clinical trials.

Performance of Salamanca refinement of the T3-AJCC8 versus the Brigham and Women's Hospital and Tübingen alternative staging systems for high-risk cutaneous squamous cell carcinoma

The Brigham and Women's Hospital and the Tübingen cutaneous squamous cell carcinoma (SCC) stratification systems propose different criteria from the American Joint Committee on Cancer, eighth edition. Our group identified prognostic subgroups within T3 stage according to the American Joint Committee on Cancer eighth edition's classification, the most common classification for high-risk cutaneous SCCs. To compare the performance and prognostic accuracy of these staging systems in a subset of high-risk cutaneous SCCs. Homogeneity, monotonicity, and McNemar tests for pairwise comparisons were assessed. Distinctiveness and relative risk of poor outcome were calculated by stage. Prognostic accuracy was compared with respect to quality (Akaike and Bayesian information criteria), concordance (Harrell C-index and Gönen and Heller concordance probability estimate), and predictive accuracy (sensitivity, specificity, negative predictive value, positive predictive value, and global accuracy). The Brigham and Women's Hospital and Salamanca systems were more distinctive, homogeneous, and monotonic than the Tübingen system. The Tübingen system was the most specific, whereas the Salamanca and Brigham and Women's Hospital systems were more sensitive. Negative predictive value was high in all 3 systems, but positive predictive value and accuracy were low overall. Alternative staging systems may partially overcome the heterogeneity and low prognostic accuracy of the American Joint Committee on Cancer, eighth edition and enable high-risk cutaneous SCCs to be stratified more reliably, but their prognostic accuracy is still low. Considering the accumulation of risk factors may improve high-risk cutaneous SCC risk stratification.

Preoperative nomogram predicting 12-year probability of metastatic renal cancer – evaluation in a contemporary cohort

ObjectivesPreoperative models, based on patient and tumor characteristics, predict risk for adverse outcomes after nephrectomy. Changes in renal tumor characteristics over the last decades, warrant further evaluation using contemporary cohorts.We aimed to validate a previously published preoperative nomogram predicting 12-year metastasis-free probability after nephrectomy for localized renal tumors in a contemporary cohort. Patients and methodsAfter obtaining institutional review board approval, data of 1,760 patients who underwent nephrectomy for a localized renal mass between 2005 and 2011 were reviewed. Preoperative images were evaluated for the presence of tumor necrosis, lymphadenopathy, and tumor size.The study outcome was metastatic-free probability. Model discrimination was assessed with Gönen and Heller's concordance probability estimate, and calibration was evaluated. ResultsThe cohort included 1,102 male and 658 female patients with a median age of 60 years. Most patients presented incidentally (84%). On imaging, 3% had evidence of lymphadenopathy, 55% had necrosis and median tumor diameter was 3.7 cm (interquartile range [IQR]: 2.5, 5.5).Median follow-up in non-metastatic patients was 7.7 years (IQR: 5.3, 9.7). Estimated 12-year metastatic-free probability was 88% (86%–90%). The model showed strong discrimination (concordance probability estimate [CPE]: 0.77), and fair calibration. The time-dependent receiver operating characteristic (ROC) curves showed strong discrimination at all-time points and the area under the curve (AUC) for year 12 was 0.83 (95% Confidence Interval: 0.78–0.89). ConclusionsWe validated the preoperative nomogram of 12-year metastasis-free probability in a contemporary cohort despite different tumor characteristics.Future studies should evaluate the role of preoperative risk stratification in patient selection for neoadjuvant treatment.

Optimal method for measuring invasive size that predicts survival in invasive mucinous adenocarcinoma of the lung.

The purpose of this study was to determine the optimal method for measuring pathological invasive size that predicts prognosis in invasive mucinous adenocarcinoma (IMA). We analyzed patients who underwent complete surgical resection for lung IMA. The invasive size of IMA was measured using two methods: (1) excluding lepidic method (ELM), that is, lepidic component was excluded from the invasive area regardless of alveolar mucin and (2) including lepidic method (ILM), that is, lepidic component was included as invasive area if alveolar space was filled with mucin. The prognostic predictability of ELM and ILM on survival was assessed using univariable and multivariable Cox regression models. The discriminative power was assessed using concordance probability estimate (CPE) and Akaike's information criteria (AIC), and the prognostic impact of the newly redefined pathological stage according to ELM or ILM was also assessed. A total of 101 patients were included. The median invasive size via ELM and ILM was 1.4cm (range, 0.0-7.7cm) and 2.1cm (range, 0.0-14.2cm), respectively. ELM had better discriminative power than ILM (ELM, HR = 1.38, AIC = 110.19, CPE = 0.671; ILM, HR = 1.19, AIC = 111.52, CPE = 0.655). Although the survival curves based on ILM crossed between T3 and T4, the overall survival (OS) curves based on ELM were sufficiently distinct from one another. ELM has higher discriminative power for OS, and thus the optimal method for measuring the pathological invasive size of IMA should exclude the lepidic component regardless of alveolar mucin.

A23 A Genomically Adjusted Clinicopathologic Model Predicts Recurrence in Resected Early-Stage Lung Squamous Cell Carcinoma

In contrast to lung adenocarcinoma, identification of clinically relevant genomic perturbations in lung squamous cell carcinoma (LUSC) remains poorly characterized. Prognostic and therapeutic decisions following surgery in early and locoregionally advanced LUSC are almost exclusively driven by the TNM classification system, omitting high-risk clinicopathologic and tumor genomic information. To address this knowledge gap, we sought to determine if a combined clinicopathologic and genomic model could predict disease-free survival (DFS) better than traditional TNM assessments in completely resected LUSC. A retrospective cohort study of a prospectively maintained database was performed for patients (N=95) with pathologic stage I-III LUSC who underwent complete resection from 2008-2018. Patients who received any induction therapy (N=9) were excluded. All patients had complete clinicopathologic data with broad-panel next-generation sequencing of the primary tumor, including matched controls to bioinformatically filter germline variants. DFS, the primary endpoint, and overall survival (OS) were estimated using Kaplan-Meier. Genomic pathway alterations (N=10) were determined as we previously reported. The prognostic model established associations between high-risk clinicopathologic variables and genomic alterations with DFS through Cox regression models. Concordance probability estimate (CPE) was used to discriminate performance between the existing TNM model and the developed prediction models. The median age was 70 years (range, 55-84), one-third were female (n=35), 55% (N=52) were pathologic stage I, and 98% (N=93) were ever-smokers. Median follow-up was 2 years. Recurrence occurred in 20% (19/95) and DFS was 74% (95% CI, 84-96%) at two years. Clinicopathologic features associated with DFS were lymphovascular invasion, visceral-pleural invasion, and pathologic stage. Tumor genomic analysis revealed alterations in the transcription factor BCL6 were independently associated with worse DFS (HR 5.23, 95% CI 1.73, 15.9, p=0.009), while mutations in the tumor suppressor ARID1A were associated with a worse OS (HR 2.98, 95% CI 0.91, 9.77, p=0.07). Pathway-centric analyses revealed no associations with our primary or secondary endpoints. Our prognostic clinicopathologic model outperformed the internally validated TNM model (CPE, 0.74 vs. 0.70) for prediction of DFS and our clinicopathologic model, adjusted for BCL6 genomic alterations, further improved discrimination (CPE=0.77). We show that integration of high-risk clinicopathologic and tumor genomic profiling better predicts DFS than TNM classification alone in early-stage, surgically resected LUSC – an observation that may facilitate enrichment in future adjuvant therapy clinical trials. This exploratory genomic analysis also suggests future studies investigate putative therapeutic vulnerabilities in LUSC tumors harboring BCL6 and ARD1A genomic alterations.

Quantitative assessment of tumor-infiltrating lymphocytes in mismatch repair proficient colon cancer

ABSTRACT Tumor infiltrating lymphocytes (TIL), which represent host adaptive response to the tumor, were first identified at scanning magnification to select areas with the highest counts on hematoxylin and eosin slides, quantitated per high-power field (HPF), and analyzed for association with recurrence-free survival (RFS) in 848 patients. Highest TIL in a single HPF was analyzed as a continuous and categorical variable, and optimal cutoff analysis was performed to predict RFS. Highest TIL count in a single HPF ranged from 0 to 45, and the optimal cutoff for TIL high vs TIL low was determined to be ≥ 3 vs < 3 with a concordance probability estimate of 0.74. In the entire cohort, 5-year RFS was 90.2% (95% CI = 83.7–94.2) in TIL high compared to 78.9% (95% CI = 74.1–82.9) in TIL low (log rank P < .0001). TIL remained significant in the mismatch repair-proficient (pMMR) cohort where 5-year RFS was 94.6% (95% CI = 88.3–97.5) in TIL high compared to 77.9% (95% CI = 69.2–84.4) in TIL low (P = .008). On multivariable analysis, TIL and AJCC Stage were independently associated with RFS in the pMMR cohort. Qualitatively in the pMMR cohort, RFS in Stage II TIL high patients was similar to that in Stage I patients and RFS in Stage III TIL high was similar to that in Stage II TIL low patients. Assessment of TIL in a single HPF using standard H&E slides provides important prognostic information independent of MMR status and AJCC stage.

Prognostic Utility of a Whole-blood Androgen Receptor-based Gene Signature in Metastatic Castration-resistant Prostate Cancer

BackgroundThe treatment paradigm for metastatic castration-resistant prostate cancer (mCRPC) has evolved significantly in recent years. Identifying predictive and/or prognostic biomarkers in the context of this rapidly expanding therapeutic armamentarium remains a pressing and unmet clinical need. ObjectiveTo develop a prognostic whole-blood gene signature for mCRPC patients. Design, setting, and participantsAs part of an ongoing prospective, multicentre biomarker research study (Australian Prostate Biomarker Alliance), we enrolled 115 mCRPC patients commencing chemotherapy (n = 34) or androgen receptor (AR) pathway inhibitors therapy (n = 81) and obtained pretreatment whole-blood samples in PAXgene RNA tubes. Gene expression was assessed using reverse transcription-polymerase chain reaction. Outcome measurements and statistical analysisGene transcripts correlating with overall survival (OS) at p < 0.10 in univariate Cox regression models were incorporated into a multigene signature. Kaplan-Meier survival estimates and multivariate analyses were used to assess association with clinical outcomes. Prognostic strength of the signature was estimated using a concordance probability estimate (CPE). Results and limitationsBased on univariate analysis for OS, the following genes were incorporated into a multigene signature: AR splice variant 7 (AR-V7), and three androgen-regulated genes: GRHL2, HOXB13, and FOXA1. The number of positive transcripts clearly stratified survival outcomes (median OS: not reached vs 24.8 mo vs 16.2 mo for 0, 1, and ≥2 transcripts, respectively; p = 0.0052). Notably, this multigene signature retained prognostic significance on multivariable analysis (hazard ratio, 2.1; 95% confidence interval, 1.1–4.0; p = 0.019). Moreover, CPE for this model was 0.78, indicating strong discriminative capacity. Limitations include short follow-up time. ConclusionsOur data demonstrate the prognostic utility of a novel whole-blood AR-based signature in mCRPC patients commencing contemporary systemic therapies. Our pragmatic assay requires minimal processing, can be performed in most hospital laboratories, and could represent a key prognostic tool for risk stratification in mCRPC. Patient summaryWe found that expression of certain genes associated with the androgen receptor could help determine how long men with advanced prostate cancer survive after starting modern drug therapies.

External validation of a prognostic score predicting overall survival for patients with brain metastases based on extracranial factors

The aim of our study was an external validation of the extracranial prognostic score predicting survival of patients with brain metastases receiving cranial irradiation on data from a single institution. A retrospective analysis of 524 patients with brain metastases treated with cranial radiotherapy in a single tertiary center was performed. Three predictive scores were calculated and assessed for their ability to discriminate prognostic groups: (i) The Recursive Partitioning Analysis (RPA) score (available for 524 patients); (ii) the Diagnosis-Specific Graded Prognostic Assessment (DS-GPA) score (464 patients); (iii) the extracranial score (EC-S) developed by Nieder et al. which is based on serum albumin, lactate dehydrogenase (LDH) and the number of extracranial organs involved (157 patients). Discrimination of each score was assessed by Gönen & Heller's concordance probability estimate (CPE). The calibration was checked by comparing median survival estimates of each risk group with the corresponding values of the datasets from which the scores were derived. Finally, a multivariable Cox regression model was built by using the least absolute shrinkage and selection operator on a large number of variables including all three scores. With a CPE = 0.626 ± 0.022, the EC-S had the best discriminatory power. The EC-S also appeared to be better calibrated and had the best ability to separate patients with a very poor prognosis: patients with combination of low albumin, elevated LDH and more than 1 extracranial organ with metastatic involvement had a median survival time of only 0.6 months (CI95% 0.1-1.1) and a hazard ratio for death of 6.36 (2.67-15.14) compared to patients with no extracranial metastases and normal levels of albumin and LDH. In the multivariable Cox model serum albumin, LDH, treatment modality, DS-GPA and EC-S were retained as prognostic factors. An ad hoc combination of both DS-GPA and EC-S into a new score was possible for 134 patients and indicated a slightly better discrimination (CPE = 0.636 ± 0.023) than either DS-GPA or EC-S alone. This study provides an independent validation of the prognostic EC-S which was the best prognostic model for defining the patients who obviously did not benefit from radiation therapy of brain metastases in terms of overall survival. The combination of the EC-S with the established DS-GPA score resulted in a slight increase in discriminatory ability. The new EC-GPA score needs further validation in larger patient cohorts.