To What Extent Can We Predict Prostate Cancer Lymph Node Involvement?

Abstract

In this issue of European Urology, Godoy et al from the Memorial Sloan-Kettering Cancer Center (MSKCC) validate a previously developed preoperative nomogram to predict lymph node involvement (LNI) of prostate cancer at pelvic lymph node dissection (PLND) during radical prostatectomy (RP) [1]. These earlier datawere derived from six institutions between 1985 and 2000 [2] and consisted of >5000 men, as opposedto thecurrent studydata,whichareexclusively from MSKCC during 2000–2008 and include >3000 men. The accuracy of the original nomogram was 76% (area under the curve [AUC]: 0.76) compared to a level of 74% (AUC: 0.74) using the Partin tables [2]. Bothmodels use prostate-specific antigen (PSA), biopsy Gleason score, and clinical T stage. Is this 2% difference clinically significant? There is really not a great answer other than to say it is more accurate. Perhaps it does matter if you prefer to use a tool akin to a slide rule or something like marking a box for bingo. In earnest, these investigators were pioneers in developing predictionmodels to be used in the clinic for prostate cancer management and the recent submission uses elegant statistical modeling. The comments in this editorial will not focus on the merits of a nomogram over other prediction models. This has recently been covered in this journal [3]. Moreover, I will not try to answer the question of whether any prediction model for LNI has played a meaningful role in the outcome of men with LNI. The concept of a PLND in prostate cancer and its nuanceswere recently reviewed by a multi-institutional collaboration in this journal and salient features related to this new article could be found there [4]. The original LNI nomogram predicted the absence of LNI [2]. The current study aimed to validate the original nomogram or develop a new one that pertains to the use of a standard PLND as opposed to the former, limited PLND.