Towards personalized follow-up: a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients

Annemieke Witteveen, Ingrid Vliegen, G.S. Sonke, J.M. Klaase, Maarten Joost IJzerman, Sabine Siesling

Research output: Contribution to conferencePaperAcademic

Abstract

Background The objective of this study was to develop and validate a conditional logistic regression model for the prediction of locoregional recurrence (LRR) of breast cancer. To make a translation to clinical practice a web based nomogram was made. Methods Women first diagnosed with early breast cancer (without distant metastasis or ingrowth in the chest wall or skin) between 2003-2006 were selected from the Netherlands Cancer Registry (N=37,278). In the first five years following primary breast cancer treatment 957 (3.0%) of the selected patients developed a LRR as a first event. Risk factors were determined using logistic regression and the risks were calculated per year, conditional on not being diagnosed with recurrence in the previous year. The presence of interaction and collinearity in the nomogram was assessed, as well as the discrimination by means of the area under the ROC curve and calibration by the Hosmer-Lemeshow goodness-of-fit test in deciles. Bootstrapping was used for internal validation. Data from 43 Dutch hospitals on primary tumours diagnosed between 2007-2008 was used for external validation of the performance of the nomogram (n=12,318). Results The final model included the variables grade, size, multifocality, and nodal involvement of the primary tumour, and whether patients were treated with radio-, chemo- or hormone therapy. The modelling group showed an area under the ROC curve of 0.84, 0.76, 0.70, 0.73 and 0.65 respectively per subsequent year after primary treatment. Model predictions were well calibrated. All effects in the validation group were in the same direction and the estimates in the validation group did not differ significantly from the modelling group. The results were incorporated in a web based nomogram. Conclusions This validated nomogram can be used as an instrument to aid clinical decision-making for personalized follow-up and to identify patients with a low or high risk of LRR who might benefit from a less or more intensive follow-up after breast cancer.
Original languageEnglish
Publication statusPublished - 25 Jun 2015
EventEuropean Congress of Epidemiology: Healthy Living 2015 - MECC, Maastricht, Netherlands
Duration: 25 Jun 201527 Jun 2015

Conference

ConferenceEuropean Congress of Epidemiology
CountryNetherlands
CityMaastricht
Period25/06/1527/06/15
Other25-06-2015 - 27-06-2015

Fingerprint

Nomograms
Breast Neoplasms
Recurrence
Logistic Models
ROC Curve
Area Under Curve
Neoplasms
Thoracic Wall
Radio
Netherlands
Calibration
Registries
Therapeutics
Hormones
Neoplasm Metastasis
Skin

Keywords

  • METIS-313614
  • IR-98465

Cite this

Witteveen, A., Vliegen, I., Sonke, G. S., Klaase, J. M., IJzerman, M. J., & Siesling, S. (2015). Towards personalized follow-up: a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients. Paper presented at European Congress of Epidemiology, Maastricht, Netherlands.
@conference{ba4f7b54be4b4a608302c9eaa7829c37,
title = "Towards personalized follow-up: a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients",
abstract = "Background The objective of this study was to develop and validate a conditional logistic regression model for the prediction of locoregional recurrence (LRR) of breast cancer. To make a translation to clinical practice a web based nomogram was made. Methods Women first diagnosed with early breast cancer (without distant metastasis or ingrowth in the chest wall or skin) between 2003-2006 were selected from the Netherlands Cancer Registry (N=37,278). In the first five years following primary breast cancer treatment 957 (3.0{\%}) of the selected patients developed a LRR as a first event. Risk factors were determined using logistic regression and the risks were calculated per year, conditional on not being diagnosed with recurrence in the previous year. The presence of interaction and collinearity in the nomogram was assessed, as well as the discrimination by means of the area under the ROC curve and calibration by the Hosmer-Lemeshow goodness-of-fit test in deciles. Bootstrapping was used for internal validation. Data from 43 Dutch hospitals on primary tumours diagnosed between 2007-2008 was used for external validation of the performance of the nomogram (n=12,318). Results The final model included the variables grade, size, multifocality, and nodal involvement of the primary tumour, and whether patients were treated with radio-, chemo- or hormone therapy. The modelling group showed an area under the ROC curve of 0.84, 0.76, 0.70, 0.73 and 0.65 respectively per subsequent year after primary treatment. Model predictions were well calibrated. All effects in the validation group were in the same direction and the estimates in the validation group did not differ significantly from the modelling group. The results were incorporated in a web based nomogram. Conclusions This validated nomogram can be used as an instrument to aid clinical decision-making for personalized follow-up and to identify patients with a low or high risk of LRR who might benefit from a less or more intensive follow-up after breast cancer.",
keywords = "METIS-313614, IR-98465",
author = "Annemieke Witteveen and Ingrid Vliegen and G.S. Sonke and J.M. Klaase and IJzerman, {Maarten Joost} and Sabine Siesling",
year = "2015",
month = "6",
day = "25",
language = "English",
note = "European Congress of Epidemiology : Healthy Living 2015 ; Conference date: 25-06-2015 Through 27-06-2015",

}

Witteveen, A, Vliegen, I, Sonke, GS, Klaase, JM, IJzerman, MJ & Siesling, S 2015, 'Towards personalized follow-up: a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients' Paper presented at European Congress of Epidemiology, Maastricht, Netherlands, 25/06/15 - 27/06/15, .

Towards personalized follow-up : a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients. / Witteveen, Annemieke; Vliegen, Ingrid; Sonke, G.S.; Klaase, J.M.; IJzerman, Maarten Joost; Siesling, Sabine.

2015. Paper presented at European Congress of Epidemiology, Maastricht, Netherlands.

Research output: Contribution to conferencePaperAcademic

TY - CONF

T1 - Towards personalized follow-up

T2 - a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients

AU - Witteveen, Annemieke

AU - Vliegen, Ingrid

AU - Sonke, G.S.

AU - Klaase, J.M.

AU - IJzerman, Maarten Joost

AU - Siesling, Sabine

PY - 2015/6/25

Y1 - 2015/6/25

N2 - Background The objective of this study was to develop and validate a conditional logistic regression model for the prediction of locoregional recurrence (LRR) of breast cancer. To make a translation to clinical practice a web based nomogram was made. Methods Women first diagnosed with early breast cancer (without distant metastasis or ingrowth in the chest wall or skin) between 2003-2006 were selected from the Netherlands Cancer Registry (N=37,278). In the first five years following primary breast cancer treatment 957 (3.0%) of the selected patients developed a LRR as a first event. Risk factors were determined using logistic regression and the risks were calculated per year, conditional on not being diagnosed with recurrence in the previous year. The presence of interaction and collinearity in the nomogram was assessed, as well as the discrimination by means of the area under the ROC curve and calibration by the Hosmer-Lemeshow goodness-of-fit test in deciles. Bootstrapping was used for internal validation. Data from 43 Dutch hospitals on primary tumours diagnosed between 2007-2008 was used for external validation of the performance of the nomogram (n=12,318). Results The final model included the variables grade, size, multifocality, and nodal involvement of the primary tumour, and whether patients were treated with radio-, chemo- or hormone therapy. The modelling group showed an area under the ROC curve of 0.84, 0.76, 0.70, 0.73 and 0.65 respectively per subsequent year after primary treatment. Model predictions were well calibrated. All effects in the validation group were in the same direction and the estimates in the validation group did not differ significantly from the modelling group. The results were incorporated in a web based nomogram. Conclusions This validated nomogram can be used as an instrument to aid clinical decision-making for personalized follow-up and to identify patients with a low or high risk of LRR who might benefit from a less or more intensive follow-up after breast cancer.

AB - Background The objective of this study was to develop and validate a conditional logistic regression model for the prediction of locoregional recurrence (LRR) of breast cancer. To make a translation to clinical practice a web based nomogram was made. Methods Women first diagnosed with early breast cancer (without distant metastasis or ingrowth in the chest wall or skin) between 2003-2006 were selected from the Netherlands Cancer Registry (N=37,278). In the first five years following primary breast cancer treatment 957 (3.0%) of the selected patients developed a LRR as a first event. Risk factors were determined using logistic regression and the risks were calculated per year, conditional on not being diagnosed with recurrence in the previous year. The presence of interaction and collinearity in the nomogram was assessed, as well as the discrimination by means of the area under the ROC curve and calibration by the Hosmer-Lemeshow goodness-of-fit test in deciles. Bootstrapping was used for internal validation. Data from 43 Dutch hospitals on primary tumours diagnosed between 2007-2008 was used for external validation of the performance of the nomogram (n=12,318). Results The final model included the variables grade, size, multifocality, and nodal involvement of the primary tumour, and whether patients were treated with radio-, chemo- or hormone therapy. The modelling group showed an area under the ROC curve of 0.84, 0.76, 0.70, 0.73 and 0.65 respectively per subsequent year after primary treatment. Model predictions were well calibrated. All effects in the validation group were in the same direction and the estimates in the validation group did not differ significantly from the modelling group. The results were incorporated in a web based nomogram. Conclusions This validated nomogram can be used as an instrument to aid clinical decision-making for personalized follow-up and to identify patients with a low or high risk of LRR who might benefit from a less or more intensive follow-up after breast cancer.

KW - METIS-313614

KW - IR-98465

M3 - Paper

ER -

Witteveen A, Vliegen I, Sonke GS, Klaase JM, IJzerman MJ, Siesling S. Towards personalized follow-up: a conditional prediction model and nomogram for risk of locoregional recurrence in early breast cancer patients. 2015. Paper presented at European Congress of Epidemiology, Maastricht, Netherlands.