Tumour architecture is an independent predictor of outcomes after nephroureterectomy: a multi-institutional analysis of 1363 patients



This article is corrected by:

  1. Errata: Corrigenda Volume 104, Issue 4, 574, Article first published online: 21 July 2009

Shahrokh F. Shariat, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, Box 27, New York, NY 10065, USA.
e-mail: sfshariat@gmail.com or shariats@mskcc.org



To assess whether tumour architecture can help to refine the prognosis of patients treated with nephroureterectomy (NU) for urothelial carcinoma (UC) of the upper urinary tract (UT), as the prognostic value of tumour architecture (papillary vs sessile) in UTUC remains elusive.


The study included 1363 patients with UTUC and treated with radical NU at 12 centres worldwide. All slides were re-reviewed according to strict criteria by genitourinary pathologists who were unaware of the findings of the original pathology slides and clinical outcomes. Gross tumour architecture was categorized as sessile vs papillary.


Papillary growth was identified in 983 patients (72.2%) and sessile growth in 380 (27.8%). The sessile growth pattern was associated with higher tumour grade, more advanced stage, lymphovascular invasion, and metastasis to lymph nodes (all P < 0.001). In multivariable Cox regression analyses that adjusted for the effects of pathological stage, grade and lymph node status, tumour architecture (sessile or papillary) was an independent predictor of cancer recurrence (hazard ratio 1.5, P = 0.002) and cancer-specific mortality (1.6, P = 0.001). Adding tumour architecture increased the predictive accuracy of a model that comprised pathological stage, grade and lymph node status for predicting cancer recurrence and cancer-specific death by a minimal but statistically significant margin (gain in predictive accuracy 1% and 0.5%, both P < 0.001).


The tumour architecture of UTUC is associated with established features of biologically aggressive disease, and more importantly, with prognosis after radical NU. Including tumour architecture in predictive models for disease progression should be considered, aiming to identify patients who might benefit from early systemic therapeutic intervention.