Survival Tree Research Articles

Prediction model for survival of younger patients with breast cancer using the breast cancer public staging database

Breast cancer (BC) is a major contributor to female mortality worldwide, particularly in young women with aggressive tumors. Despite the need for accurate prognosis in this demographic, existing studies primarily focus on broader age groups, often using the SEER database, which has limitations in variable selection. This study aimed to develop an ML-based model to predict survival outcomes in young BC patients using the BC public staging database. A total of 3,401 patients with BC were included in the study. Patients were categorized as younger (n = 1574) and older (n = 1827). We applied several survival models—Random Survival Forest, Gradient Boosting Survival, Extra Survival Trees (EST), and penalized Cox models (Lasso and ElasticNet)—to compare mortality characteristics. The EST model outperformed others in predicting mortality for both age groups. Older patients exhibited a higher prevalence of comorbidities compared to younger patients. Tumor stage was the primary variable used to train the model for mortality prediction in both groups. COPD was a significant variable only in younger patients with BC. Other variables exhibited varying degrees of consistency in each group. These findings can help identify high-risk young female patients with BC who require aggressive treatment by predicting the risk of mortality.

Machine learning for predicting device-associated infection and 30-day survival outcomes after invasive device procedure in intensive care unit patients

This study aimed to preliminarily develop machine learning (ML) models capable of predicting the risk of device-associated infection and 30-day outcomes following invasive device procedures in intensive care unit (ICU) patients. The study utilized data from 8574 ICU patients who underwent invasive procedures, sourced from the Medical Information Mart for Intensive Care (MIMIC)-IV version 2.2 database. Patients were allocated into training and validation datasets in a 7:3 ratio. Seven ML models were employed for predicting device-associated infections, while five models were used for predicting 30-day survival outcomes. Model performance was primarily evaluated using the receiver operating characteristic (ROC) curve for infection prediction and the survival model’s concordance index (C-index). Top-performing models progressively reduced the number of variables based on their importance, thereby optimizing practical utility. The inclusion of all variables demonstrated that extreme gradient boosting (XGBoost) and extra survival trees (EST) models yielded superior discriminatory performance. Notably, when restricted to the top 10 variables, both models maintained performance levels comparable to when all variables were included. In the validation cohort, the XGBoost model, with the top 10 variables, achieved an area under the curve (AUC) of 0.810 (95% CI 0.808–0.812), an area under the precision-recall curve (AUPRC) of 0.226 (95% CI 0.222–0.230), and a Brier score (BS) of 0.053 (95% CI 0.053–0.054). The EST model, with the top 10 variables, reported a C-index of 0.756 (95% CI 0.754–0.757), a time-dependent AUC of 0.759 (95% CI 0.763–0.775), and an integrated Brier score (IBS) of 0.087 (95% CI 0.087–0.087). Both models are accessible via a web application. The internally evaluated XGBoost and EST models demonstrated exceptional predictive accuracy for device-associated infection risks and 30-day survival outcomes post-invasive procedures in ICU patients. Further validation is required to confirm the clinical utility of these two models in future studies.

Optimizing Prognostic Predictions in Liver Cancer with Machine Learning and Survival Analysis.

This study harnesses RNA sequencing data from the Cancer Genome Atlas to unearth pivotal genetic markers linked to the progression of liver hepatocellular carcinoma (LIHC), a major contributor to cancer-related deaths worldwide, characterized by a dire prognosis and limited treatment avenues. We employ advanced feature selection techniques, including sure independence screening (SIS) combined with the least absolute shrinkage and selection operator (Lasso), smoothly clipped absolute deviation (SCAD), information gain (IG), and permutation variable importance (VIMP) methods, to effectively navigate the challenges posed by ultra-high-dimensional data. Through these methods, we identify critical genes like MED8 as significant markers for LIHC. These markers are further analyzed using advanced survival analysis models, including the Cox proportional hazards model, survival tree, and random survival forests. Our findings reveal that SIS-Lasso demonstrates strong predictive accuracy, particularly in combination with the Cox proportional hazards model. However, when coupled with the random survival forests method, the SIS-VIMP approach achieves the highest overall performance. This comprehensive approach not only enhances the prediction of LIHC outcomes but also provides valuable insights into the genetic mechanisms underlying the disease, thereby paving the way for personalized treatment strategies and advancing the field of cancer genomics.

Glycemic control in diabetic patients improved overall lung cancer survival across diverse populations.

The consequence of diabetes on lung cancer overall survival (OS) is debated. This retrospective study used 2 large lung cancer databases to assess comprehensively diabetes effects on lung cancer OS in diverse demographic populations, including health disparity. The University of Texas MD Anderson Cancer Center database (32 643 lung cancer patients with 11 973 patients with diabetes) was extracted from electronic health records (EHRs) using natural language processing (NLP). Associations were between diabetes and lung cancer prognostic features (age, sex, race, body mass index [BMI], insurance status, smoking, stage, and histopathology). Hemoglobin A1C (HgbA1c) and glucose levels assessed glycemic control. Validation was with a Louisiana cohort (17 768 lung cancer patients with 5402 patients with diabetes) enriched for health disparity cases. Kaplan-Meier analysis, log-rank test, multivariable Cox proportional hazard models, and survival tree analyses were employed. Lung cancer patients with diabetes exhibited marginally elevated OS or no statistically significant difference versus nondiabetic patients. When examining OS for 2 glycemic levels (HgbA1c > 7.0 or glucose > 154 mg/dL vs HgbA1c > 9.0 or glucose > 215 mg/dL), a statistically significant improvement in OS occurred in lung cancer patients with controlled versus uncontrolled glycemia (P < .0001). This improvement spanned sex, age, smoking status, insurance status, stage, race, BMI, histopathology, and therapy. Survival tree analysis revealed that obese and morbidly obese patients with controlled glycemia had higher lung cancer OS than comparison groups. These findings indicate a need for optimal glycemic control to improve lung cancer OS in diverse populations with diabetes.

Physical activity modifies cognitive impairment-associated mortality risks among chronic kidney disease

BackgroundOlder chronic kidney disease (CKD) patients frequently face unrecognized cognitive impairment and excess mortality. Physical activity (PA) reduces cognitive decline but whether PA modifies cognitive impairment-associated mortality remains unknown. MethodsFrom 2005 to 2011, 30,561 older Taiwanese CKD patients were enrolled. Patients were divided into intact cognition (≥8 scores), mild (6–7 scores), and severe (≤5 scores) cognitive impairment groups by the Short Portable Mental Status Questionnaire (SPMSQ), and were also categorized into high-PA (≥60 min/week of moderate-intensity PA), low-PA (20–60 min/week) or inactive (<20 min/week) groups. Cox regression was conducted to evaluate the individual and joint associations of cognitive impairment and PA on all-cause and cardiovascular mortality. ResultsAfter a median follow-up of 4.52 years, the all-cause mortality were higher in CKD patients with severe (multivariable-adjusted hazard ratio [aHR] 2.31; 95% confidence interval [CI] 2.05–2.60) and mild (aHR 1.74; CI 1.51–1.99) cognitive impairment than cognitively intact ones. Remarkably, decreased PA amount interacted and amplified the cognitive impairment-associated mortality risks. Notably, the high-PA status linked to lower overall mortality risks both in mild (aHR 0.65; CI 0.45–0.93) and severe (aHR 0.73; CI 0.54–0.99) cognitively-impaired patients as compared to inactivity. Survival tree analysis indicated the least mortality in those with high PA and >8 SPMSQ scores. Similar associations were found in the cardiovascular mortality. LimitationsResidual confounding and single ethnicity. ConclusionsCognitive impairment defined by SPMSQ was progressively associated with higher mortality among elderly CKD. Higher PA linked to lower cognitive impairment-associated death risks, and could be promoted for longevity benefits.

Comparison of Extremely Randomized Survival Trees and Random Survival Forests: A Simulation Study

Abstract. Purpose: This simulation study investigates the Extremely Randomized Survival Trees (EST) model, a machine learning technique expected to handle survival analysis, particularly in large survival datasets, effectively. The study compares the performance of the EST model with that of the Random Survival Forest (RSF) model, focusing on the C-index value to determine which model performs better. Methods: The analysis begins with the generation of 540 simulated datasets, created by combining three levels of sample sizes, two levels of censoring proportions, three types of hazard functions, and 30 repetitions for each scenario. The simulation data were split into 80% training and 20% testing data. The training data were used to build the EST and RSF models, while the test data were used to evaluate their performance. The model with the highest C-index value was deemed the best performer, as a higher C-index indicates superior model performance. Result: The results indicate that the sample size, type of hazard function, and the method used influence that model performance. The EST model significantly outperformed the RSF model when the sample size was large, though no significant difference was observed when the sample size was small or medium. Additionally, the EST model consistently demonstrated faster computation times across all simulation scenarios. Novelty: This study provides a pioneering exploration into applying decision tree algorithms, specifically EST and RSF, in survival analysis. While these methods have been extensively studied in regression and classification contexts, their application in survival analysis remains relatively unexplored.

AI hybrid survival assessment for advanced heart failure patients with renal dysfunction

Renal dysfunction (RD) often characterizes the worse course of patients with advanced heart failure (AHF). Many prognosis assessments are hindered by researcher biases, redundant predictors, and lack of clinical applicability. In this study, we enroll 1736 AHF/RD patients, including data from Henan Province Clinical Research Center for Cardiovascular Diseases (which encompasses 11 hospital subcenters), and Beth Israel Deaconess Medical Center. We developed an AI hybrid modeling framework, assembling 12 learners with different feature selection paradigms to expand modeling schemes. The optimized strategy is identified from 132 potential schemes to establish an explainable survival assessment system: AIHFLevel. The conditional inference survival tree determines a probability threshold for prognostic stratification. The evaluation confirmed the system’s robustness in discrimination, calibration, generalization, and clinical implications. AIHFLevel outperforms existing models, clinical features, and biomarkers. We also launch an open and user-friendly website www.hf-ai-survival.com, empowering healthcare professionals with enhanced tools for continuous risk monitoring and precise risk profiling.

Development and validation of a stromal-immune signature to predict prognosis in intrahepatic cholangiocarcinoma.

Intrahepatic cholangiocarcinoma (ICC) is a highly desmoplastic tumor with poor prognosis even after curative resection. We investigated the associations between the composition of the ICC stroma and immune cell infiltration and aimed to develop a stromal-immune signature to predict prognosis in surgically treated ICC. We recruited 359 ICC patients and performed immunohistochemistry to detect α-smooth muscle actin (α-SMA), CD3, CD4, CD8, Foxp3, CD68, and CD66b. Aniline was used to stain collagen deposition. Survival analyses were performed to detect prognostic values of these markers. Recursive partitioning for a discrete-time survival tree was applied to define a stromal-immune signature with distinct prognostic value. We delineated an integrated stromal-immune signature based on immune cell subpopulations and stromal composition to distinguish subgroups with different recurrence-free survival (RFS) and overall survival (OS) time. We defined four major patterns of ICC stroma composition according to the distributions of α-SMA and collagen: dormant (α-SMAlow/collagenhigh), fibrogenic (α-SMAhigh/collagenhigh), inert (α-SMAlow/collagenlow), and fibrolytic (α-SMAhigh/collagenlow). The stroma types were characterized by distinct patterns of infiltration by immune cells. We divided patients into six classes. Class I, characterized by high CD8 expression and dormant stroma, displayed the longest RFS and OS, whereas Class VI, characterized by low CD8 expression and high CD66b expression, displayed the shortest RFS and OS. The integrated stromal-immune signature was consolidated in a validation cohort. We developed and validated a stromal-immune signature to predict prognosis in surgically treated ICC. These findings provide new insights into the stromal-immune response to ICC.

PSA doubling time 4.65 months as an optimal cut-off of Japanese nonmetastatic castration-resistant prostate cancer

A multicenter study of nonmetastatic castration-resistant prostate cancer (nmCRPC) was conducted to identify the optimal cut-off value of prostate-specific antigen (PSA) doubling time (PSADT) that correlated with the prognosis in Japanese nmCRPC. Of the 515 patients diagnosed and treated for nmCRPC at 25 participating Japanese Urological Oncology Group centers, 450 patients with complete clinical information were included. The prognostic values of clinical factors were evaluated with respect to prostate specific antigen progression-free (PFS), cancer-specific survival (CSS), and overall survival (OS). The optimal cutoff value of PSADT was identified using survival tree analysis by Python. The Median PSA and PSADT at diagnosis of nmCRPC were 3.3 ng/ml, and 5.2 months, respectively. Patients treated with novel hormonal therapy (NHT) showed significantly longer PFS (HR: hazard ratio 0.38, p < 0.0001) and PFS2 (HR 0.45, p < 0.0001) than those treated with vintage nonsteroidal antiandrogen agent (Vintage). The survival tree identified 4.65 months as the most prognostic PSADT cutoff point. Among the clinical and pathological factors PSADT of < 4.65 months remained an independent prognostic factor for OS (HR 2.96, p = 0.0003) and CSS (HR 3.66, p < 0.0001). Current data represented optimal cut-off of PSADT 4.65 months for a Japanese nmCRPC.

Survival Prognostication in Patients with Differentiated Thyroid Cancer and Distant Metastases: A SEER Population-Based Study.

Background: For patients with thyroid cancer, distant metastasis is a significant predictor of poor outcome. Since distant metastasis occurs in less than 10% of patients with differentiated thyroid cancer, correlates of survival in this vulnerable patient population remain understudied. This study aimed to identify prognostic groups among patients with differentiated thyroid cancer and distant metastases and to determine the role of, and interactions between, patient and tumor characteristics in determining survival. Methods: We identified adult patients diagnosed with differentiated thyroid cancer with distant metastases from the U.S. SEER-17 cancer registry (2010-2019). Analyses were performed using Cox proportional hazards regression, survival trees, and random survival forest. Relative importance of patient and tumor factors important for disease-specific and overall survival was assessed based on the random survival forest analyses. Results: Cohort consisted of 2411 patients with differentiated thyroid cancer with distant metastases followed for a median of 62 months. Most common histopathologic subtype (86.0%) was papillary thyroid cancer, and the most common sites of distant metastasis were the lungs (33.7%) and bone (18.9%). Cox proportional hazards model illustrated significant associations between survival and the following: patient age (p < 0.001), tumor size (p < 0.01), and site of distant metastasis (p < 0.05). Survival tree analyses identified three distinct prognostic groups based on disease-specific survival (DSS) (5-year survival of the prognostic groups was 92%, 64%, and 41%; p < 0.001) and four distinct prognostic groups based on overall survival (OS) (5-year survival of the prognostic groups was 96%, 84%, 57%, and 31%; p < 0.001). The first split in the survival trees for DSS and OS was by age at diagnosis (≤57 years vs. ≥58 years) with subsequent splits based on presence/absence of lung metastases, tumor size (≤4 cm vs. >4 cm), and patient age. A total of 558 patients (23.1%) died from thyroid cancer, and 757 patients (31.4%) died from all causes during the study period. Conclusions: This study identifies distinct prognostic groups for patients with differentiated thyroid cancer with distant metastases and highlights the importance of patient age, lung metastases, and tumor size for determining both disease-specific and overall survival. These findings inform risk stratification and treatment decision-making in this understudied patient population.

Poster 387: Application of Survival Machine Learning Methodology to Predict Time-Dependent Risk of Revision ACLR Graft Failure in the MARS Cohort

Objectives: Previous machine learning (ML) analysis has shown that a novel ML model built using the Multicenter Anterior Cruciate Ligament (ACL) Revision Study (MARS) cohort data can predict graft failure at 6 years postoperatively as a binary outcome with moderate ability. However, binary classification does not consider the time-to-event component of graft failure as an outcome. In contrast, survival models can factor the time-to-event component of an outcome, incorporate time-dependent variables, and appropriately process censored data to deliver more clinically meaningful outcome predictions. While the previous binary classifier can provide information about whether the primary outcome graft failure will occur at 6 years, survival analysis models are able to provide information on when graft failure occurs (for example, within 2, 4, or 6 years). Therefore, this study sought to apply ML survival analysis methodology to the MARS cohort data to produce a model that can most accurately estimate the probability of revision ACL reconstruction (rACLR) graft failure within 2, 4, and 6 years postoperatively, and identify factors that impact probability estimates at each of these time horizons. Methods: rACLR patients were prospectively recruited by the MARS group. Preoperative radiographs, surgeon-reported intraoperative findings, and 2 and 6-year follow-up data on patient-reported outcomes (PROs), subsequent knee surgeries, and graft failure were obtained. ML models including Cox Proportional Hazards (CoxPH), XGBoost Survival Embeddings (XGBoost), Random Survival Forest (RSF), Extra Survival Tree (EST), and CoxBoost, were built to predict the probability of graft failure within 2, 4, and 6 years postoperatively. Validated performance metrics and feature importance measures were used to evaluate model performance. Results: The cohort included 1142 patients, of which 4.8% (n = 55) experienced documented graft failure during the 6-year postoperative period. CoxBoost demonstrated the highest discriminative power across all studied time points (Timepoint: Time Dependent C-index | 2-year: 0.725 ± 0.127 | 4-year: 0.669 ± 0.057 | 6-year: 0.645 ± 0.054), with well-calibrated scores (Timepoint: Time dependent Brier Score | 2-year: 0.023 ± 0.012 | 4-year: 0.035 ± 0.010 | 6-year: 0.046 ± 0.012) as listed in Table 1 and Table 2 respectively. Based on the Kaplan Meier survival plot, the survival probability (probability of not tearing the ACL graft) at 2, 4, 6 years was 0.977, 0.965, and 0.954 respectively. Within the context of the predictive models, features deemed important for CoxBoost differed from those deemed important for CoxPH predictive ability. Conclusions: Survival machine learning models can predict the risk of rACLR graft failure up to 6 years postoperatively, with the most robust predictions at the 2-year postoperative timepoint. While external validation with further combined registry datasets is required, this present study builds on prior ML analyses towards the development of a bedside calculator for rACLR outcome prediction and risk stratification.

Effects of stand structure on ecological benefits and tree growth of Haloxylon ammodendron plantations: A meta‐analysis

AbstractHaloxylon ammodendron is a typical tree species in arid region for windbreak and sand fixation. Understanding how stand structure (i.e., afforestation density, forest type, and stand age) of H. ammodendron impacts ecological benefits and tree growth status has great implications for desertification control in arid area. We obtained a dataset of 446 observations from 79 studies related to H. ammodendron plantations for meta‐analysis. Stand age was the key factor affecting performances of windbreak, tree canopy, and soil properties while afforestation density was more important in determining tree survival rate. Wind speed in H. ammodendron plantations was reduced by 52% in general compared to that at the site without any plantations and decreased as the increasing density. With the increase of stand age, soil carbon, nitrogen, and phosphorus accumulated continuously, but soil moisture in 0–60 cm depth decrease when trees grew up to nearly 20 years. Within 4000 plants ha−1, responses of survival rate, tree height, ground diameter, and canopy decreased with the increase of afforestation density. Compared with mono‐species plantation, mixed afforestation was more beneficial to windbreak and survival, while it damaged tree growth. Combining the ecological functions and tree growth of H. ammodendron plantations, the afforestation density and forest age should be controlled to 450–900 plants ha−1 and 20 years, respectively, as the upper limit. Therefore, besides reasonable afforestation density, management measures for regulations of forest age and existing stand density were important to maintain higher ecological benefits and better tree growth of H. ammodendron plantations.

Advancing Solar Energetic Particle Event Prediction through Survival Analysis and Cloud Computing. I. Kaplan–Meier Estimation and Cox Proportional Hazards Modeling

Solar energetic particles (SEPs) pose significant challenges to technology, astronaut health, and space missions. This initial paper in our two-part series undertakes a comprehensive analysis of the time to detection for SEPs, applying advanced statistical techniques and cloud-computing resources to deepen our understanding of SEP event probabilities over time. We employ a range of models encompassing nonparametric, semiparametric, and parametric approaches, such as the Kaplan–Meier estimator and Cox Proportional Hazards models. These are complemented by various distribution models—including exponential, Weibull, lognormal, and log-logistic distributions—to effectively tackle the challenges associated with “censored data,” a common issue in survival analysis. Employing Amazon Web Services and Python’s “lifelines” and “scikit-survival” libraries, we efficiently preprocess and analyze large data sets. This methodical approach not only enhances our current analysis, but also sets a robust statistical foundation for the development of predictive models, which will be the focus of the subsequent paper. In identifying the key determinants that affect the timing of SEP detection, we establish the vital features that will inform the machine-learning (ML) techniques explored in the second paper. There, we will utilize advanced ML models—such as survival trees and random survival forests—to evolve SEP event prediction capabilities. This research is committed to advancing space weather, strengthening the safety of space-borne technology, and safeguarding astronaut health.

Water level impact on pine seedlings in greenhouse conditions: assessing growth and survival potential in ditched and managed peatlands

ABSTRACT Tree establishment on peatlands has various adverse effects on the environment, with one of the most significant being their transformation from carbon sinks to carbon sources. This transformation has largely been instigated by economic-driven ditching initiatives. In this study, 80 peat-rooted pine seedlings were subjected to hydrological scenarios corresponding to natural, ditched, and rewetted conditions to investigate how different management strategies affect tree growth and survival. The study was conducted in a greenhouse where all plants were exposed to identical conditions except for the water level, and focused on factors like stomatal conductance, plant survival, length, biomass, and radial tree growth. Wet conditions, specifically treatments rewetted and natural, resulted in consistently lower stomatal conductance compared to drier treatments. Plant survival was affected, with 15 deaths in the rewetted and 2 in natural groups. Moreover, length, biomass, radial growth, and cell formation were significantly lower for the groups exposed to wet conditions. Rewetting can therefore effectively control tree colonisations, and thereby preventing water consumption, litter fertilisation, and other positive feedback effects for the trees that might be negative for the carbon uptake and biodiversity in peatlands. This study thereby offers valuable insights for rewetting initiatives in tree colonised peatland ecosystems.

Machine learning with electrocardiograms to optimize mortality risk stratification in patients with suspected acute coronary syndrome

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Heart, Lung, and Blood Institute (NHLBI) and National Center for Advancing Translational Sciences (NCATS). Background The importance of risk stratification in patients with suspected acute coronary syndrome (ACS) extends beyond diagnosis and immediate treatment. It influences the precision of care delivery and the allocation of resources to those at highest risk of adverse events. Purpose We sought to evaluate the prognostic value of electrocardiogram (ECG) feature-based machine learning models to risk-stratify long-term mortality in those with suspected ACS. Methods This was a multicenter prospective observational cohort study of consecutive, non-traumatic patients evaluated at the emergency department for suspected ACS. The derivation cohort included 4,015 patients from a University Medical Center (age 59±16 years, 47% women, 80% training with 10-fold cross-validation and 20% testing). Ascertainment of all-cause death was based on numerous data sources, including the Centers for Disease Control and Prevention’s National Death Index registry. We trained six machine learning methods for survival analysis using 73 morphological ECG features from presenting prehospital 10-second 12-lead ECGs. We used predicted risk scores in a variational Bayesian Gaussian mixture model to define three risk groups corresponding to low, moderate, and high-risk and compared the resulting classification performance against the HEART score. The external testing cohort included 3,095 patients from a University (age 59±15 years, 44% women). Results The mortality rate was 20.3% in the derivation cohort after an average follow-up period of 3.53 years (IQR 1.75 - 5.32). Extra Survival Trees outperformed other forecasting models during cross-validation. In the internal testing cohort, the derived risk groups were significantly predictive of survival (Kaplan-Meier log-rank test statistic = 121.14, p&amp;lt;0.001), outperforming the classification performance using the HEART score (Figure 1). Our machine learning-based risk stratification model detected &amp;gt;90% of death events missed by low-risk HEART score (Figure 2). Ruling-out the low-risk group, it achieved a negative predictive value of 93.4% with a sensitivity of 85.9% (compared to 89.0% and 75.0%, respectively, for the HEART score). This classification performance generalized well to our external testing cohort. Compared to those in the low-risk group, patients at moderate (OR = 3.62 [1.35-9.74]) and high (OR = 6.12 [2.38-15.75]) risk had significantly higher odds of 30-day cardiovascular death (n=109, 3.5%). Conclusions Using only features from the 10-second 12-lead ECG, we derived and externally validated a machine learning model that stratifies the mortality risk of patients during long-term follow up. This model outperformed current standard of care based on the HEART score. Future work should validate the potential of this decision-support tool in influencing treatment plans and resource allocation.Figure 1.Clustering performanceFigure 2.Sankey diagram

Abstract PO5-17-03: Relative Dose Intensity and Protraction of Neoadjuvant Chemotherapy in Early-Stage and Locally Advanced Breast Cancer: A collaborative multicenter analysis in Colombia

Abstract Introduction: Breast cancer (BC) is the most frequent neoplasm in Colombia. In contrast to high-income countries, the expected 5-year overall survival (OS) remains below 80% over the last years. Neoadjuvant chemotherapy (NACT) has reduced the risk of relapse and death by achieving higher pathological complete response (ypCR) rates, but delays and dose reductions could impact prognosis negatively. We have previously reported the outcomes of NACT from a real-world multicenter historical cohort of Colombian early and locally advanced BC patients. We aimed to analyze NACTRelative Dose Intensity (RDI) and protraction data as related to clinical outcomes in the same cohort. Methods: We performed a secondary analysis of the primary database of 312 BC patients treated with NACT where ypCR rate was 34·6%, with 88·2% 5-year OS. We considered RDI and protraction as independent variables for the same outcomes. The RDI was estimated using the Hrynuik method, and protraction was defined as the difference between planned and actual completion time for the NACT exceeding 7 days from standard 21-day cycles. Chemotherapy regimens are based on the National Hematology and Oncology Association (ACHO) guidelines. We excluded cases with missing or conflicting data. Univariate, and bivariate analysis, and partition survival trees were performed to examine variable interactions. Results: From the original cohort of 312 patients, 47 and 7 additional patients were excluded due to missing protraction and RDI data, respectively; the final analysis included 261 and 254 patients, respectively. The average days to complete planned chemotherapy cycles were 154,4 (SD 32,29 days). 203 patients (77,8%), 26 patients (10%), and 11 patients (4,2%) received complete 8,7 and 6 NACT cycles, respectively. Overall, 175 patients (67%) had protracted chemotherapy cycles, being highest in the 7-cycle subgroup (77%). 169 patients (66,5%) had a dose intensity over 85% and 90 patients (35.43%) achieved an RDI of 100% or more. The mean RDI for the cohort was 79,8% (SD 19,91%) and it differed according to expression profile subgroups in decreasing order as follows: HER2 pure 98,85% (SD 19,59%), Luminal A 96,03% (SD 21,28%), Triple Negative 97,81% (SD 19,73%), Luminal B HER+ and HER2- being lowest with 90,43% (SD 19,47%) and 87,75% (SD 23,22%), respectively. Neither NACT RDI nor protraction was associated significantly with ypCR, conservative surgery, or all-cause OS. Although statistically not-significant, sensitivity analysis through different RDI thresholds (20%-100%) demonstrated improved discriminatory capacity over ypCR for RDI &amp;gt;= 85%. Conclusion: On average, NACT-RDI greater than 85% was observed in Colombian patients with early and locally advanced BC similar to other cohorts. However, it was not associated with ypCR, conservative surgery, or OS. Also, NACT protraction beyond 7 days did not adversely affect clinical outcomes. However, this conclusion is limited due to the retrospective nature and loss of real-world data. Adverse events and their impact on RDI were not analyzed either. Further research is warranted to understand RDI's impact on clinical outcomes, and a large-scale trial is proposed to validate these findings. Citation Format: Alejandro Murillo, Andres Acevedo, Santiago Betancur, Mariana Borras-Osorio, Isabel Munevar, William Mantilla, Luis Pino. Relative Dose Intensity and Protraction of Neoadjuvant Chemotherapy in Early-Stage and Locally Advanced Breast Cancer: A collaborative multicenter analysis in Colombia [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-17-03.

The International Association for the Study of Lung Cancer Pleural Mesothelioma Staging Project: Proposal for Revision of the TNM Stage Groupings in the Forthcoming (Ninth) Edition of the TNM Classification for Pleural Mesothelioma

IntroductionThe eighth edition of the TNM classification of pleural mesothelioma (PM) saw substantial changes in T and N components and stage groupings. The International Association for the Study of Lung Cancer collected data into a multinational database to further refine this classification. This ninth edition proposal incorporates changes proposed in the clinical (c)T component but not the pathologic T component, to include size criteria, and further refines TNM stage groupings for PM. MethodsData were submitted through electronic data capture or batch transfer from institutional databases. Survival was measured from diagnosis date. Candidate stage groups were developed using a recursive partitioning and amalgamation algorithm applied to all cM0 cases for clinical stage and subsequently for pathologic stage. Cox models were developed to estimate survival for each stage group. ResultsOf 3598 submitted cases, 2192 were analyzable for overall clinical stage and 445 for overall pathologic stage. Recursive partitioning and amalgamation generated survival tree on overall survival outcomes restricted to cM0, with newly proposed (ninth edition) cT and cN component-derived optimal stage groupings of stage I (T1N0), II (T1N1; T2N0), IIIA (T1N2; T2N1/2; any T3), IIIB (any T4), and IV (any M1). Although cT and pathologic T descriptors are different in the ninth edition, aligning pathologic stage groupings with clinical stage produced better discrimination than did retaining eighth edition pathologic stage groupings. ConclusionsTo our knowledge, this revision of the clinical TNM classification for PM is the first to incorporate the measurement-based proposed changes in cT category. The pathologic TNM aligns with clinical TNM.

Random forests for survival data: which methods work best and under what conditions?

Few systematic comparisons of methods for constructing survival trees and forests exist in the literature. Importantly, when the goal is to predict a survival time or estimate a survival function, the optimal choice of method is unclear. We use an extensive simulation study to systematically investigate various factors that influence survival forest performance - forest construction method, censoring, sample size, distribution of the response, structure of the linear predictor, and presence of correlated or noisy covariates. In particular, we study 11 methods that have recently been proposed in the literature and identify 6 top performers. We find that all the factors that we investigate have significant impact on the methods' relative accuracy of point predictions of survival times and survival function estimates. We use our results to make recommendations for which methods to use in a given context and offer explanations for the observed differences in relative performance.

Explainable and visualizable machine learning models to predict biochemical recurrence of prostate cancer.

Machine learning (ML) models presented an excellent performance in the prognosis prediction. However, the black box characteristic of ML models limited the clinical applications. Here, we aimed to establish explainable and visualizable ML models to predict biochemical recurrence (BCR) of prostate cancer (PCa). A total of 647 PCa patients were retrospectively evaluated. Clinical parameters were identified using LASSO regression. Then, cohort was split into training and validation datasets with a ratio of 0.75:0.25 andBCR-related features were included in Cox regression and five ML algorithm to construct BCR prediction models. The clinical utility of each model wasevaluated by concordance index (C-index) values and decision curve analyses (DCA). Besides, Shapley Additive Explanation (SHAP) values were used to explain the features in the models. We identified 11 BCR-related features using LASSO regression, then establishing five ML-based models, including random survival forest (RSF), survival support vector machine (SSVM), survival Tree (sTree), gradient boosting decision tree (GBDT), extreme gradient boosting (XGBoost), and a Cox regression model, C-index were 0.846 (95%CI 0.796-0.894), 0.774 (95%CI 0.712-0.834), 0.757 (95%CI 0.694-0.818), 0.820 (95%CI 0.765-0.869), 0.793 (95%CI 0.735-0.852), and 0.807 (95%CI 0.753-0.858), respectively. The DCA showed that RSF model had significant advantages over all models. In interpretability of ML models, the SHAP value demonstrated the tangible contribution of each feature in RSF model. Our score system provide reference for the identification for BCR, and the crafting of a framework for making therapeutic decisions for PCa on a personalized basis.

A Novel Decision Tree Model for Predicting the Cancer-Specific Survival of Patients with Bladder Cancer Treated with Radical Cystectomy.

Background/Objectives: The aim was to develop a decision tree and a new prognostic tool to predict cancer-specific survival in patients with urothelial bladder cancer treated with radical cystectomy. Methods: A total of 11,834 patients with bladder cancer treated with radical cystectomy between 2004 and 2019 from the SEER database were randomly split into the derivation (n = 7889) and validation cohorts (n = 3945). Survival curves were estimated using conditional decision tree analysis. We used Multiple Imputation by Chained Equations for the treatment of missing values and the pec package to compare the predictive performance. We extracted data from our model following CHARMS and assessed the risk of bias and applicability with PROBAST. Results: A total of 4824 (41%) patients died during the follow-up period due to bladder cancer. A decision tree was made and 12 groups were obtained. Patients with a higher AJCC stage and older age have a worse prognosis. The risk groups were summarized into high, intermediate and low risk. The integrated Brier scores between 0 and 191 months for the bootstrap estimates of the prediction error are the lowest for our conditional survival tree (0.189). The model showed a low risk of bias and low concern about applicability. The results must be externally validated. Conclusions: Decision tree analysis is a useful tool with significant discrimination. With this tool, we were able to stratify patients into 12 subgroups and 3 risk groups with a low risk of bias and low concern about applicability.