Conditional survival represents the adjustment in the probability of survival that is defined according to the duration of event-free follow-up. Since it dynamically accounts for changes in the risk of death over time, conditional survival may be more specific than unadjusted survival, especially in patients with event-free follow-up of several years. In this study, we tested conditional cancer-specific mortality (CSM)-free estimates in non-metastatic bladder cancer patients treated with radical cystectomy. Unlike other previous studies,1-3 in the current study4 we assessed the risk of five-year conditional CSM-free estimates accounting for the combined effect of T and N stages.
Within the SEER database (2004-2015), conditional five-year CSM-free estimates were calculated according to T and N stage groupings (pTa/pTis/pT1N0 vs. pT2N0 vs. pT3N0 vs. pT4N0 vs. pTanyN1-3), Specifically, CS was calculated as the probability of survival for x additional years, given y years of accumulated survival. Then, separate multivariable Cox regression (MCR) models predicting CSM were fitted to examine the possible variation for risk of CSM over time. We also developed a conditional nomogram predicting five-year CSM, according to the duration of event-free interval.
We identified 13,595 eligible patients. Stage distribution was as follows: 7.9% pTa/pTis/pT1N0, 37.2% pT2N0, 21.1% pT3N0, 8.9% pT4N0 and 24.9% pTanyN1-3. A direct relationship between event-free follow-up duration and CSM-free probability was recorded in all stage groupings. However, the magnitude of survival gains according to event-free follow-up was the highest in patients with the most aggressive stage groupings. For example, in pTanyN1-3 patients, conditional five-year CSM-free estimates increased from 37.5% to 84.0% after five years of event-free follow-up (survival gain of 23.6%). Conversely, in pTa/pTis/pT1N0 patients, the conditional five-year CSM-free estimates were virtually unchanged from 90.1% to 91.8% (survival gain of +1.7%). Nonetheless, we also recorded important attrition rates, in part due to CSM, that were the highest in pTanyN1-3 patients, where only 12.9% of patients were still alive after five years of event-free follow-up.
The graphical representation of HRs over time showed a decrease to non-statistically significant levels at 48 months of event free follow-up for pT2N0 and pT3N0, as well as at 60 months for pT4N0 and pTanyN1-3, relative to pTa/pTis/pT1aN0. These data suggest that few if any additional CSM events may be expected in pT2N0 and pT3N0 individuals with event-free follow-up of 36 months or more. Similarly, in pT4N0 and pTanyN1-3 patients few if any additional CSM events may be expected after event-free follow-up of 48 months or more.
Finally, the developed nomogram with its conditional counterpart showed a moderate discrimination capacity (c-index of 0.69). Its prediction was predominantly based on T stage and to a lesser extent to N stage, with a marginal contribution by other variables. For these reasons, the nomogram cannot be endorsed for routine use in clinical practice. Nonetheless, we do believe that the graphical representation of HRs for specific combinations of T and N stages at specific event-free follow-up intervals provides more valuable tools for CS predictions after specific event-free follow-up intervals and should be incorporated into patient counseling.
In conclusion, from a clinical perspective, information derived from the current conditional model could be combined with standard pathological stage information obtained at radical cystectomy to tailor the frequency of imaging and/or tailor the administration of adjuvant therapy.
Written by: Carlotta Palumbo, MD, and Pierre I. Karakiewicz, MD, Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Quebec, Canada. Twitter: @CPalumbo87, @sophieknipper, @AngelaPecoraro3, @DrShariat, @aleantonellibs1, @pikarakiewicz,
References:
1. Sun M, Abdollah F, Bianchi M, et al. Conditional survival of patients with urothelial carcinoma of the urinary bladder treated with radical cystectomy. Eur J Cancer 2012;48:1503–11. doi:10.1016/j.ejca.2011.11.024
2. Ploussard G, Shariat SF, Dragomir A, et al. Conditional Survival After Radical Cystectomy for Bladder Cancer: Evidence for a Patient Changing Risk Profile over Time. Eur Urol 2014;66:361–70. doi:10.1016/j.eururo.2013.09.050.
3. Kang M, Kim HS, Jeong CW, et al. Prognostic factors for conditional survival in patients with muscle-invasive urothelial carcinoma of the bladder treated with radical cystectomy. Sci Rep 2015;5. doi:10.1038/srep12171.
4. Palumbo C, Mistretta FA, Knipper S, et al. How cancer-specific mortality changes over time after radical cystectomy: Conditional survival of patients with nonmetastatic urothelial carcinoma of the urinary bladder. Urol Oncol. 2019 Jul 8. doi: 10.1016/j.urolonc.2019.05.020.
Read the Abstract