Carsten Nieder1, Minesh P Mehta2, Hans Geinitz3, Anca L Grosu4. 1. Department of Oncology and Palliative Medicine, Nordland Hospital, 8092 Bodø, Norway; Department of Clinical Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway. Electronic address: carsten.nieder@nlsh.no. 2. Department of Radiation Oncology, Miami Cancer Institute, Miami, FL, USA. 3. Department of Radiation Oncology, St. Vincent's Hospital, Linz, Austria. 4. Department of Radiation Oncology, University Medical Center, Medical Faculty, University of Freiburg, Germany; German Cancer Consortium (DKTK) Partner Site Freiburg, Germany.
Abstract
OBJECTIVE: To review published nomograms that predict endpoints such as overall survival (OS) or risk of intracranial relapse in patients with brain metastases from solid tumors. METHODS: The methods and results of nomogram studies identified by a systematic search were extracted and compared, stratified by endpoint predicted by the respective nomograms. In particular, validation strategies (external/internal), concordance indices (cut-off 0.75) and comparisons to older models were analyzed. RESULTS: Six publications reported on prediction of OS. Most of these analyses focused on one particular primary tumor site, e.g., breast cancer or hepatocellular carcinoma, while the largest study included different primary tumor sites. The median number of patients was 244. Three of six studies included external validation cohorts. With few exceptions, concordance indices <0.75 were reported. In all studies reporting this endpoint, the nomogram outperformed older prognostic scores. Two nomograms focused on development of new brain metastases after radiosurgery (one externally validated), one on survival free from salvage whole brain radiotherapy (WBRT) after radiosurgery, and one on neurologic and non-neurologic death in patients receiving radiosurgery after WBRT failure. All concordance indices of these 4 nomograms were <0.70. CONCLUSION: Taking into account concordance indices and comparisons to older prognostic models, the most promising, externally validated nomograms are the breast cancer and the non-small cell lung cancer nomogram predicting OS, and the distant brain failure after radiosurgery nomogram. Additional validation studies as well as continuous monitoring of the models' performance appear necessary to ensure their clinical applicability in the present era of rapidly changing treatment paradigms.
OBJECTIVE: To review published nomograms that predict endpoints such as overall survival (OS) or risk of intracranial relapse in patients with brain metastases from solid tumors. METHODS: The methods and results of nomogram studies identified by a systematic search were extracted and compared, stratified by endpoint predicted by the respective nomograms. In particular, validation strategies (external/internal), concordance indices (cut-off 0.75) and comparisons to older models were analyzed. RESULTS: Six publications reported on prediction of OS. Most of these analyses focused on one particular primary tumor site, e.g., breast cancer or hepatocellular carcinoma, while the largest study included different primary tumor sites. The median number of patients was 244. Three of six studies included external validation cohorts. With few exceptions, concordance indices <0.75 were reported. In all studies reporting this endpoint, the nomogram outperformed older prognostic scores. Two nomograms focused on development of new brain metastases after radiosurgery (one externally validated), one on survival free from salvage whole brain radiotherapy (WBRT) after radiosurgery, and one on neurologic and non-neurologic death in patients receiving radiosurgery after WBRT failure. All concordance indices of these 4 nomograms were <0.70. CONCLUSION: Taking into account concordance indices and comparisons to older prognostic models, the most promising, externally validated nomograms are the breast cancer and the non-small cell lung cancer nomogram predicting OS, and the distant brain failure after radiosurgery nomogram. Additional validation studies as well as continuous monitoring of the models' performance appear necessary to ensure their clinical applicability in the present era of rapidly changing treatment paradigms.
Authors: Jonathan W Rick; Maryam Shahin; Ankush Chandra; Cecilia Dalle Ore; John K Yue; Alan Nguyen; Garima Yagnik; Soumya Sagar; Saman Arfaie; Manish K Aghi Journal: Crit Rev Oncol Hematol Date: 2019-07-22 Impact factor: 6.312