BERKELEY, CA (UroToday.com) - Dr. Guilherme Godoy and colleagues from Memorial Sloan-Kettering Cancer Center report a new nomogram for predicting lymph node involvement (LNI) in prostate cancer (CaP) patients.
Their work appears on the online edition of European Urology.
The investigators had previously reported a nomogram to predict the risk of LNI at the time of radical prostatectomy (RP). The accuracy of such a tool, they point out, is in part related to the extent of lymphadenectomy (LND) and tissue processing. The present study was executed to assess the discrimination of the nomogram (meaning the ability to rank patients by their risk) and calibration of the nomogram (meaning how well the predicted incidence of LNI matches the observed incidence). The study objective was to rebuild the nomogram coefficients with only patients who had undergone the standard LND for either validation or updating to reflect the contemporary environment.
The dataset included 3,721 patients who underwent RP between 2000 and 2008 by either open or minimally invasive approaches. Clinical and pathological variables were assessed. A median number of 11 lymph nodes were removed, and 60% of study subjects had at least 10 nodes excised. The mean predicted probability of LNI was 3.2% and 5.7% in the previously published 3- and 4-variable nomograms. The 3-variable nomogram included PSA, clinical stage, and Gleason grade; the 4-variable also included the categorical variable of incidence of LNI (ranging from 1.5% to 7%). The respective AUCs for these 3- and 4-variable nomograms were 0.861 and 0.859, indicating strong discriminative ability. The new nomogram also showed high discriminative ability with an AUC = 0.862. Of current interest is identification of patients with a <20% chance of LNI and perhaps to avoid a LND in this cohort. Both of the previously reported nomograms were poorly calibrated for these patients, but the newly calibrated nomogram was much better. For patients with a predicted probability ≥2% (high risk) vs. <2% (low risk) for LNI, the prior 3-variable nomogram and new nomogram were discordant in only 4.9% of patients. In other words, only 2 of 165 patients reclassified by the new nomogram as low-risk had LNI and 19 patients reclassified as high-risk by the new nomogram had negative lymph nodes. The prior 4-variable nomogram and new nomogram were discordant in 47% of patients; all of 1,752 patients were reclassified as low-risk by the new nomogram. Compared with the prior 4-variable nomogram, the new one correctly reclassified all but 11 of the 1,752 men. Thus, the new nomogram has the highest clinical net benefit and is best suited to identify patients’ risk for LNI. While the discriminative ability was similar for all 3 nomograms, the new one has better calibration.
Godoy G, Chong KT, Cronin A, Vickers A, Laudone V, Touijer K, Guillonneau B, Eastham JA, Scardino PT, Coleman JA
Eur Urol. 2011 Jan 18. Epub ahead of print.