Display Settings:

Format

Send to:

Choose Destination
Clin Cancer Res. 2009 Feb 1;15(3):1013-8. doi: 10.1158/1078-0432.CCR-08-1888.

A simple and accurate model for prediction of cancer-specific mortality in patients treated with surgery for primary penile squamous cell carcinoma.

Author information

  • 1Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, QC, Canada.

Abstract

PURPOSE:

Cancer-specific mortality (CSM) of patients with primary penile squamous cell carcinoma (PPSCC) may be quite variable. Recently, a nomogram was developed to provide standardized and individualized mortality predictions. Unfortunately, it relies on a large number (n = 8) of specific variables that are unavailable in routine clinical practice. We attempted to develop a simpler prediction rule with at least equal accuracy in predicting CSM after surgical removal of PPSCC.

EXPERIMENTAL DESIGN:

The predictive rule was developed on a cohort of 856 patients identified in the 1988 to 2004 Surveillance, Epidemiology and End Results (SEER) database. The predictors consisted of age, race, SEER stage (localized versus regional versus metastatic), tumor grade, type of surgery (excisional biopsy, partial penectomy, and radical penectomy), and of lymph node status (pN0 versus pN1-3 versus pNx). A look-up table based on Cox regression model-derived coefficients was used for prediction of 5-year CSM. The predictive rule accuracy was tested using the Harrell's modification of the area under the receiver operating characteristics curve.

RESULTS:

SEER stage and histologic grade achieved independent predictor status and qualified for inclusion in the model. The model achieved 73.8% accuracy for prediction of CSM at 5 years after surgery. Both predictors achieved independent predictor status in competing risk regression models addressing CSM, where other cause mortality was controlled for.

CONCLUSION:

Despite equivalent accuracy, our predictive rule predicting 5-year CSM in patients with PPSCC is substantially less complex (2 versus 8 variables) than the previously published model.

PMID:
19188173
[PubMed - indexed for MEDLINE]
Free full text
PubMed Commons home

PubMed Commons

0 comments
How to join PubMed Commons

    Supplemental Content

    Full text links

    Icon for HighWire
    Loading ...
    Write to the Help Desk