Risk Factors for Local Relapse and Inferior Disease-free Survival After Breast-conserving Management of Breast Cancer: Recursive Partitioning Analysis of 2161 Patients

Hammer, J.; Geinitz, Hans; Nieder, Carsten; Track, C.; Seewald, D.H.; Petzer, Andreas; Helfgott, Ruth; Spiegl, K.J.; Heck, Dietmar; Bräutigam, E.

doi:10.1016/j.clbc.2018.08.001

Cited by 9 publications

(9 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If a significant association was found ( p < 0.05), the respective parameter was included in a multinominal logistic regression analysis. Finally, recursive partitioning analysis (RPA), a well-established method in brain metastases research (a technique of building decision trees) [21], previously utilized by our group [22] was employed. A treatment utilization rate of 75% was defined as an outcome of interest (an arbitrary, percentile-based decision).…”

Section: Methodsmentioning

confidence: 99%

Recursive Partitioning Analysis of Systemic Therapy after Radiotherapy in Patients with Brain Metastases

2021

Self Cite

View full text Add to dashboard Cite

Purpose: The purpose of this study was to identify factors associated with the initiation or continuation of systemic treatment after brain irradiation. The outcome of interest was a utilization rate of at least 75%, given that active extracranial disease is common in patients with brain metastases. If left untreated, extracranial disease limits survival, regardless of successful local treatment of the brain metastases. In this context, systemic therapy has been shown to improve survival, e.g., after whole-brain radiotherapy. Patients and Methods: The study included 185 patients with active extracranial disease, 60% of whom received systemic therapy. Results: Survival from the start of brain irradiation was longest in patients who received additional immune checkpoint inhibitors, endocrine treatment, or anti-HER-2 drugs. After uni- and multivariate analyses, Eastern Cooperative Oncology Group performance status (PS) was selected as the first prediction criterion in the recursive partitioning analysis (RPA) decision tree analysis. RPA was successful for patients with PS 0–1, but patients with PS 2 had lower treatment utilization rates (maximum 60–70%, with a disease-dependent impact of age and LabBM score [blood test results]). The highest utilization rates were observed in (1) patients with PS 0 and (2) those with breast cancer, small-cell lung cancer, or lung adenocarcinoma with PS 1. Conclusions: These results inform the multidisciplinary discussion and treatment planning for the common scenario of simultaneous intra- and extracranial metastases.

show abstract

Section: Methodsmentioning

confidence: 99%

Recursive Partitioning Analysis of Systemic Therapy after Radiotherapy in Patients with Brain Metastases

2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…Five studies calculated or provided rationale for their sample size for model development and were all based on flawed methodology [82]. This included, one study used 10 events per variable when developing a logistic regression model and a neural network [49], and another study used estimation of a relative hazard ratio between prognostic groups to calculate their sample size [83]. Two studies considered their sample size restricted by the size and availability of the existing data they were using (one randomised controlled trial [84] and one cohort study [66]) and one study justified sample size based on a time interval (e.g., consecutive adult patients over a 2-year period to allow a sufficient sample size for randomization to the training and validation data sets) [54].…”

Section: Candidate Predictors and Sample Sizementioning

confidence: 99%

“…Four studies used a combination of approaches; one study used split sample and bootstrapping [95], two studies used split sample and cross-validation [64,96], and one study used cross-validation and bootstrapping [97]. For 11 studies, internal validation methods were unclear (18%) [65,70,75,80,83,84,[98][99][100][101][102].…”

Section: Validation Proceduresmentioning

confidence: 99%

“…Only 14 studies explicitly mentioned methods to handle loss to follow-up (n = 14/62 studies, 17%), of which six studies excluded patients that were lost to follow up [63,77,86,89,98,107], and one study reported that the 'definition of treatment failure does not capture patients lost to follow-up due to future treatments at other institutions or due to the cessation of treatment for other reasons' [56]. For the remaining seven studies, patients who were lost of follow up were included in the study and outcome definition [53,65,67,83,90,100,109]. For example, Hammer et al reported that 'if no event of interest had occurred, patients were censored at the time of last documented contact with the hospital' [83].…”

Section: Missing Data and Censoringmentioning

confidence: 99%

“…For the remaining seven studies, patients who were lost of follow up were included in the study and outcome definition [53,65,67,83,90,100,109]. For example, Hammer et al reported that 'if no event of interest had occurred, patients were censored at the time of last documented contact with the hospital' [83].…”

Section: Missing Data and Censoringmentioning

confidence: 99%

See 2 more Smart Citations

Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review

Dhiman

Navarro

et al. 2022

BMC Med Res Methodol

View full text Add to dashboard Cite

Background Describe and evaluate the methodological conduct of prognostic prediction models developed using machine learning methods in oncology. Methods We conducted a systematic review in MEDLINE and Embase between 01/01/2019 and 05/09/2019, for studies developing a prognostic prediction model using machine learning methods in oncology. We used the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement, Prediction model Risk Of Bias ASsessment Tool (PROBAST) and CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) to assess the methodological conduct of included publications. Results were summarised by modelling type: regression-, non-regression-based and ensemble machine learning models. Results Sixty-two publications met inclusion criteria developing 152 models across all publications. Forty-two models were regression-based, 71 were non-regression-based and 39 were ensemble models. A median of 647 individuals (IQR: 203 to 4059) and 195 events (IQR: 38 to 1269) were used for model development, and 553 individuals (IQR: 69 to 3069) and 50 events (IQR: 17.5 to 326.5) for model validation. A higher number of events per predictor was used for developing regression-based models (median: 8, IQR: 7.1 to 23.5), compared to alternative machine learning (median: 3.4, IQR: 1.1 to 19.1) and ensemble models (median: 1.7, IQR: 1.1 to 6). Sample size was rarely justified (n = 5/62; 8%). Some or all continuous predictors were categorised before modelling in 24 studies (39%). 46% (n = 24/62) of models reporting predictor selection before modelling used univariable analyses, and common method across all modelling types. Ten out of 24 models for time-to-event outcomes accounted for censoring (42%). A split sample approach was the most popular method for internal validation (n = 25/62, 40%). Calibration was reported in 11 studies. Less than half of models were reported or made available. Conclusions The methodological conduct of machine learning based clinical prediction models is poor. Guidance is urgently needed, with increased awareness and education of minimum prediction modelling standards. Particular focus is needed on sample size estimation, development and validation analysis methods, and ensuring the model is available for independent validation, to improve quality of machine learning based clinical prediction models.

show abstract

Risk factors for lymph node metastasis in rectal neuroendocrine tumors: A recursive partitioning analysis based on multicenter data

Wang

Zhang

Lin

et al. 2021

Journal of Surgical Oncology

View full text Add to dashboard Cite

show abstract

Risk Factors for Local Relapse and Inferior Disease-free Survival After Breast-conserving Management of Breast Cancer: Recursive Partitioning Analysis of 2161 Patients

Cited by 9 publications

References 34 publications

Recursive Partitioning Analysis of Systemic Therapy after Radiotherapy in Patients with Brain Metastases

Recursive Partitioning Analysis of Systemic Therapy after Radiotherapy in Patients with Brain Metastases

Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review

Risk factors for lymph node metastasis in rectal neuroendocrine tumors: A recursive partitioning analysis based on multicenter data

Contact Info

Product

Resources

About