Premium
Development and external validation of a prognostic tool for prediction of cancer‐specific mortality after complete loco‐regional pathological staging for squamous cell carcinoma of the penis
Author(s) -
Sun Maxine,
Djajadiningrat Rosa S.,
Alnajjar Hussain M.,
Trinh QuocDien,
Graafland Niels M.,
Watkin Nick,
Karakiewicz Pierre I.,
Horenblas Simon
Publication year - 2015
Publication title -
bju international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.773
H-Index - 148
eISSN - 1464-410X
pISSN - 1464-4096
DOI - 10.1111/bju.12677
Subject(s) - medicine , nomogram , stage (stratigraphy) , pathological , radiology , proportional hazards model , penis , oncology , lymphovascular invasion , dissection (medical) , cancer , penile cancer , metastasis , surgery , paleontology , biology
Objective To develop a novel postoperative prognostic tool, which attempts to integrate both pathological tumour stage and histopathological factors, for prediction of cancer‐specific mortality ( CSM ) of squamous cell carcinoma of the penis ( SCCP ). Patients and Methods Patients with SCCP treated with inguinal lymph node dissection ( ILND ) or sentinel LN biopsy at a single institution were used for nomogram development and internal validation ( n = 434), while a second cohort was used for external validation ( n = 338). Multivariable C ox proportional hazards were used to examine the prognostic ability of patient age, a modified tumour staging that distinguishes between spongiosum and cavernosum body ingrowth tumours, a modified LN staging that integrates information on presence/absence of LN metastasis, extent of inguinal LN metastases, pelvic LN involvement, and extranodal involvement, and tumour grade. Model performance was quantified using measures of discrimination and calibration. Results Overall, 36% of patients had positive LN metastases ( n = 156). In univariable analyses, the modified tumour and LN staging systems were statistically significantly associated with CSM , and remained in the final model with a discrimination of 89% within internal validation, and 95% within external validation. Calibration was nearly perfect. Conclusions The newly developed model integrates important prognostic factors, which existing models do not consider. Its performance was highly accurate using measures of discrimination and calibration.