Multivariable modeling of radiotherapy outcomes, including dose–volume and clinical factors

Naqa, Issam El; Bradley, Jeffrey D.; Blanco, Angel I.; Lindsay, Patricia; Vićić, M.; Hope, Andrew; Deasy, Joseph O.

doi:10.1016/j.ijrobp.2005.11.022

Cited by 154 publications

(143 citation statements)

References 32 publications

Supporting

Mentioning

140

Contrasting

Order By: Relevance

“…[10] As described elsewhere, outcomes modeling is performed as a two-step process: in the first step we study correlation vs. prediction of best fit models as a function of the number of allowed parameters. [8] We compute bootstrap and leave-one-out predictive performance as a function of allowed parameters (usually similar, but not always). The model is fit for data included and tested against data excluded.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

A Nomogram to Predict Radiation Pneumonitis, Derived From a Combined Analysis of RTOG 9311 and Institutional Data

Bradley

Hope

Naqa

et al. 2007

International Journal of Radiation Oncology*Biology*Physics

Self Cite

163

109

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

“…All modeled dose, volume, and positional factors were extracted using CERR. Model determination was performed using an open-source software system we have developed called DREES (dose response explorer system) [8] …”

Section: Introductionmentioning

confidence: 99%

A Nomogram to Predict Radiation Pneumonitis, Derived From a Combined Analysis of RTOG 9311 and Institutional Data

Bradley

Hope

Naqa

et al. 2007

International Journal of Radiation Oncology*Biology*Physics

Self Cite

163

109

View full text Add to dashboard Cite

show abstract

“…The bootstrapping technique was applied for logistic regression variable selection. 42 A similar methodology was applied in the study by Cella et al 18 The authors reported a three-variable NTCP model for Hodgkin's lymphoma. Bakhshandeh et al 28 (65 patients, 11 patients with NPC) compared 6 RHT NTCP models, and the mean dose model was identified as the best model according to Akaike's information criterion value.…”

Section: Boomsma Et Almentioning

confidence: 99%

Nomogram for radiation-induced hypothyroidism prediction in nasopharyngeal carcinoma after treatment

Luo

et al. 2017

The British Journal of Radiology

View full text Add to dashboard Cite

The authors Ren Luo and Mei Li contributed equally to this work.Objective: The aim of this study was to develop a nomogram for radiation-induced hypothyroidism (RHT) prediction. Methods: We collected data from 164 patients with nasopharyngeal carcinoma (NPC) in our previous prospective study. Biochemical hypothyroidism was defined as a serum thyroid-stimulating hormone level greater than the normal value. We collected both clinical and dose-volume factors. A univariate Cox regression analysis was performed to identify RHT risk factors. Optimal predictors were selected according to the least absolute shrinkage and selection operator (LASSO). We then selected the Cox regression models that best balanced the prediction performance and practicability to build a nomogram for RHT prediction. Results: There were 38 (23.2%) patients who developed RHT, and the median follow-up was 24 months. The univariate Cox regression analysis indicated that gender, minimum dose, mean dose (D mean ) and V 25 -V 60 [V x (%), the percentage of thyroid volume receiving .x Gy] of the thyroid were significantly associated with RHT. The variables of gender, receiving chemotherapy or not (chemo), D mean and V 50 were selected using the LASSO analysis. A nomogram based on a three-variable (gender, chemo and V 50 ) Cox regression model was constructed, and its concordance index was 0.72. Good accordance between prediction and observation was showed by calibration curves in the probability of RHT at 18, 24 and 30 months. Conclusion: This study built a nomogram for RHT in NPC survivors by analyzing both clinical and dose-volume parameters using LASSO. Thus, the individual dose constraint could be achieved in a visual format. Advances in knowledge: This study used LASSO to more accurately address the multicollinear problem between variables. The resulting nomogram will help physicians predict RHT. INTRODUCTIONRadiation-induced hypothyroidism (RHT) is a common late complication after irradiation to the neck area. The majority of early RHT cases are silent because subclinical hypothyroidism (HT) can be detected only by thyroid hormone tests [elevated serum thyroid-stimulating hormone (TSH) and normal serum free thyroxine (fT4) and/ or serum free triiodothyronine (fT3)]. However, other cases show overt disease [overt hypothyroidism (overt HT)] that presents with elevated TSH and decreased fT4 and/or fT3. The incidence of subclinical HT is 24-50%, and overt HT develops in 6-20% of patients with head and neck cancer who receive radiotherapy (RT).1 HT has been demonstrated to be correlated with the development of cardiovascular and pulmonary disorders in addition to diabetes mellitus, 2-7 which increases the adverse effects of RHT on patient quality of life.

show abstract

“…The same data were also analyzed using the DREES software, 10 which utilizes a sequential forward algorithm for variable selection, 12 along with a leave-one-out crossvalidation technique. At each model order, models are ranked according to the frequency in which they were selected as the best model in the cross-validation process.…”

Section: Iic Patient Datamentioning

confidence: 99%

“…In DREES, the variable selection is performed using sequential forward selection, and the robustness of the model is verified using cross-validation ͑CV͒ techniques such as the bootstrap test and the leave-one-out ͑LOO͒ method. 12 An alternative approach to the variable selection problem is via the utilization of a genetic algorithm. 13,14 Genetic algorithms ͑GAs͒ are numerical optimization algorithms inspired by both natural selection and natural genetics.…”

Section: Introductionmentioning

confidence: 99%

A genetic algorithm for variable selection in logistic regression analysis of radiotherapy treatment outcomes

et al. 2008

View full text Add to dashboard Cite

A given outcome of radiotherapy treatment can be modeled by analyzing its correlation with a combination of dosimetric, physiological, biological, and clinical factors, through a logistic regression fit of a large patient population. The quality of the fit is measured by the combination of the predictive power of this particular set of factors and the statistical significance of the individual factors in the model. We developed a genetic algorithm ͑GA͒, in which a small sample of all the possible combinations of variables are fitted to the patient data. New models are derived from the best models, through crossover and mutation operations, and are in turn fitted. The process is repeated until the sample converges to the combination of factors that best predicts the outcome. The GA was tested on a data set that investigated the incidence of lung injury in NSCLC patients treated with 3DCRT. The GA identified a model with two variables as the best predictor of radiation pneumonitis: the V30 ͑p = 0.048͒ and the ongoing use of tobacco at the time of referral ͑p = 0.074͒. This two-variable model was confirmed as the best model by analyzing all possible combinations of factors. In conclusion, genetic algorithms provide a reliable and fast way to select significant factors in logistic regression analysis of large clinical studies.

show abstract

Multivariable modeling of radiotherapy outcomes, including dose–volume and clinical factors

Cited by 154 publications

References 32 publications

A Nomogram to Predict Radiation Pneumonitis, Derived From a Combined Analysis of RTOG 9311 and Institutional Data

A Nomogram to Predict Radiation Pneumonitis, Derived From a Combined Analysis of RTOG 9311 and Institutional Data

Nomogram for radiation-induced hypothyroidism prediction in nasopharyngeal carcinoma after treatment

A genetic algorithm for variable selection in logistic regression analysis of radiotherapy treatment outcomes

Contact Info

Product

Resources

About