EURURO Vol. 72 No. 4

484 656

Over the last decades, several multimodal bladder-sparing
protocols using radiation therapy (RT) have been tested to
challenge the well-established dogma of radical cystectomy
(RC) as the only effective local treatment option for muscle-
invasive urothelial carcinoma of the bladder (UCB). The
most-studied conservative approach remains trimodal
therapy (TMT), comprising maximal transurethral resection
of bladder tumor (TURBT) followed by 60 Gy of RT with
concurrent radiosensitizing chemotherapy delivered in a
split or continuous course
However, existing evidence is inconclusive with regard
to the comparative oncological outcomes associated with
delivery of TMT versus RC. Indeed, a unique randomized
controlled trial failed to meet the original recruitment
targets and very few retrospective studies compared
TMT to RC, with their own set of limitations . Against
this backdrop, we hypothesized that these two local
treatment options may have a similar effect on overall
survival (OS). We used the National Cancer Data Base to
assess the comparative effectiveness of TMT versus RC in a
large sample of contemporary US patients with muscle-
invasive UCB.
From a population of 341 667 men and women
diagnosed with bladder cancer between 2004 and
2011(ICD-0-3 codes C67.0–C67.9), we identified 12 843 in-
dividuals treated with TMT or RC for localized muscle-
invasive UCB (cT2–4N0M0) who were considered in our
final study population (Supplementary Fig. 1). The TMT
group included those who received TURBT followed by 60–
65 Gy of RT delivered to the bladder with concurrent
single- or multiple-agent radiosensitizing chemotherapy,
as well as those who underwent immediate salvage RC
after 39 Gy of chemoradiation. Patients who underwent
RC with or without perioperative chemotherapy and who
did not receive any RT to the bladder before surgery were
included in the RC group.
To account for selection bias, differences observed in
baseline characteristics between the TMT and RC groups
were controlled for with inverse probability of treatment
weighting (IPTW)–adjusted analyses . Balance in cov-
ariates between treatment groups was evaluated using the
standardized differences approach and Kernel density
plots. IPTW-adjusted Kaplan-Meier curves were calculated
to compare OS between TMT and RC. The proportional
hazards assumption was tested using the Grambsch-
Therneau approach. An unbiased bootstrapping method
was subsequently used to test for difference in median OS
between treatment groups. In addition, an IPTW-adjusted
Cox regression model with a time-varying covariate
including treatment as a main effect and an interaction
term between treatment and time variables was fitted
. From the latter model, we tested for equal survival
curves using a x2 test with two degrees of freedomthat both
the main treatment effect and the interaction between
treatment and time equaled zero . In addition, a 3-mo
conditional landmark IPTW-adjusted survival analysis was
performed to assess the impact of immortal time bias on
our findings. Finally, we conducted exploratory analyses to
determine the heterogeneity of the treatment effect
according to age (continuous), gender (female vs male),
Charlson comorbidity index (CCI; 1 vs 0) and cT stage
( cT3 vs cT2) by testing interaction terms within the IPTW-
adjusted Cox model.
All statistical analyses were performed using Stata v.14.0
(StataCorp, College Station, TX, USA). Two-sided statistical
significance was defined as p < 0.05. An institutional review
board waiver was obtained before the study was conducted.
Overall, 1257 (9.8%) and 11 586 (90.2%) patients with
clinically localized muscle-invasive UCB underwent TMT
and RC, respectively (Supplementary Fig. 1). Unweighted
and weighted baseline characteristics of eligible patients,
stratified according to treatment group, are reported in
Results of multivariable logistic regression analysis
predicting receipt of TMT versus RC are reported in
Supplementary Table 1. Following IPTW adjustment, all
standardized differences were <10% (Supplementary Fig. 2).
The distribution of propensity scores demonstrated ade-
quate balance between the treatment groups(Supplemen-
tary Fig. 3A,B), indicating that the treatment groups were
subsequently comparable.
The median follow-up was 44 mo (interquartile range
27–63) in patients alive at last follow-up, while 6627
(51.6%) deaths from any cause occurred over the study
period. The proportional hazards assumption was rejected
(p < 0.001). IPTW-adjusted Kaplan-Meier curves A)
showed that median OS was similar between TMT (40 mo,
95% confidence interval [CI] 34–46) and RC (43 mo, 95% CI
41–45; p = 0.3). In IPTW-adjusted Cox regression analysis
with a time-varying covariate, TMT was associated with a
significant adverse effect on OS after 25 mo of follow-up
(hazard ratio [HR] 1.37, 95% CI 1.16–1.59; p < 0.001),
whereas there was no significant difference before 25 mo of
follow-up (HR 0.93, 95% CI 0.83–1.04; p = 0.2). The 3-mo
conditional IPTW-adjusted analysis showed little impact of
immortal time bias on the short-term (HR 0.99, 95% CI 0.89–
1.12; p = 0.9) and long-term (HR 1.37, 95% CI 1.16–1.58;
p <0.001) treatment effects ).
Interaction terms indicated that the adverse treatment
effect of TMT versus RC decreased significantly with age (HR
0.99, 95% CI 0.98–0.99; p = 0.003), while no significant
interaction was observed with gender (HR 0.93, 95% CI
0.74–1.18; p = 0.6), CCI (HR 0.84, 95% CI 0.69–1.03; p = 0.1),
or cT stage (HR 0.97, 95% CI 0.77–1.22; p = 0.8).
Corroborating data from prospective series separately
evaluating the oncological outcomes of TMT and RC
for muscle-invasive UCB, our study revealed no significant
difference in median OS between the treatment groups.
Moreover, TMT and RC demonstrated similar OS before 2 yr
of follow-up. However, after 2 yr, individuals treated with
TMT were 1.40-fold more likely to die following presenta-
tion with localized muscle-invasive UCB. These findings
suggest that the potential long-term benefit of RC may be
attenuated by the immediate risk of postoperative mortali-
ty. Finally, with regard to patient selection, we observed
that the benefit of RC was less pronounced in older patients
who may not live long enough to benefit from such
treatment, whereas no significant interaction was found
with gender, CCI, and cT stage. [9_TD$DIFF]That said, these results
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 4 8 3 – 4 8 7 484