EURURO Vol. 73 No. 1

34 148

1. Introduction
There is widespread agreement in the academic community
that men with low-risk prostate cancer should be managed
by active surveillance. The approach is recommended in
guidelines based on evidence that risk of prostate cancer
mortality is very low and not improved by treatment
Yet resistance to active surveillance remains, with one
recent study demonstrating that while rates of active
surveillance have increased, a majority of menwith low-risk
prostate cancer are nonetheless treated curatively
A key rationale for conservative management is the
avoidance of treatment-related morbidities, such as erectile
dysfunction. However, an argument can be made that the
possibility of such morbidities actually justifies immediate
surgery. This argument starts from the premise that many
men managed conservatively do end up being treated: in
one well-known study, the probability of radical treatment
was 57% by 15 yr Given that the risk of postoperative
erectile dysfunction increases with age , patients who
delay treatment may lose the window of opportunity to
recover erectile function after surgery. We have heard this
argument expressed as: “I’m probably going to have to
operate on you eventually, so I might as well treat you now
while you are young and have a good chance of recovery.”
Numerous studies have compared erectile function
between men undergoing surgery versus active surveil-
lance, the randomized ProtecT study being a notable
example . A challenge of such studies is that functional
outcomes in conservatively managed patients depend on
the rate of crossover to surgery: a more aggressive approach
to monitoring will lead to more treatment and hence poorer
functional outcomes. An alternative is to use a statistical
modeling approach, which allows the rate of surgery to be
varied. At Memorial Sloan Kettering Cancer Center, patients
undergoing radical prostatectomy (RP) complete pretreat-
ment and post-treatment patient-reported outcomes as a
routine part of clinical care. These data allow us to model a
quantitative comparison between immediate and delayed
surgery: comparing pretreatment function at different ages
gives an estimate of how erectile function would change in
an untreated man; comparing postsurgery recovery at
different ages allows us to evaluate the marginal benefit of
early surgery. Here we report a novel modeling study to
compare long-term erectile function comparing immediate
surgery versus active surveillance.
2. Materials and methods
2.1. Patients and outcomes
Following institutional review board approval, we identified
5865 patients through our prospectively maintained database who
underwent open or minimally-invasive nerve-sparing RP from an
experienced attending surgeon from 2009 through 2013, and who did
not receive adjuvant therapy. Given that we were interested in patients
potentially eligible for active surveillance, we excluded men with any
clinical features of high-risk prostate cancer, including prostate-specific
antigen 20 ng/ml, biopsy Gleason grade group 3, and clinical tumor
stage > cT2b. Men who did not undergo bilateral nerve-sparing surgery
were also omitted, leaving a cohort of 1581. Note that although we
included some patients (eg, Gleason grade group 2) who are not
considered eligible for active surveillance in some institutions, this does
not affect our findings, as functional outcomes are not importantly
different in these patients compared with those with lower risk disease.
Four hundred and fi fty men without preoperative erectile function
scores were excluded, as well as 28 men without a postoperative score
(fi nal cohort n = 1103). There were no significant differences in patient
age, preoperative prostate-specific antigen, comorbidities, number of
positive biopsy cores, surgical margin status, extracapsular extension,
seminal vesical invasion, lymph node invasion, pathology Gleason, or
pathology stage between the men with and without preoperative
erectile data (all p > 0.05).
Patient-reported erectile function was electronically collected
through our web-based platform using the International Index of
Erectile Function 6 (IIEF-6; range, 1–30). The surveys are administered at
baseline (preoperative period) and postoperatively at 3 mo, 6 mo, 9 mo,
12 mo, 18 mo, and 24 mo or shortly before any postoperative
appointments. As is standard, patients were asked to report function
on oral medication—if they took Viagra or other phosphodiesterase type
5 inhibitors—but without the use of injection therapy or other erectile
function aids.
2.2. Statistical analysis
Our primary aim was to model the effects of immediate surgery versus
active surveillance on long-term erectile function. We plotted preopera-
tive IIEF-6 scores against age to assess the natural rate of functional
decline due to aging. Reported erectile scores in the 2-yr period following
surgery were used to assess postsurgical recovery. Due to more limited
data on erectile function by age after 2 yr, we assumed that recovery
plateaus after 2 yr, and thereafter function declines with age at the same
rate as for preoperative function.
It is well known that comorbidities have an adverse effect on
erectile function. We sought to assess whether function declined at a
faster rate due to aging for men with more comorbidities. Using linear
regression models, we investigated the effect of patient age and
comorbidity status on baseline and postoperative IIEF-6 scores, with
an interaction term in the model to examine the effect of patient
comorbidities on the rate of decline of erectile function with increasing
age. Data on major comorbidities known to affect erectile function
(cardiovascular disease, hypertension requiring medical therapy,
diabetes mellitus, and peripheral vascular disease) were captured
and included in the model categorized as none, one, two, or three or
more major comorbidities. A signifi cant interaction between age and
the number of comorbidities would indicate that function declines
faster for unhealthy patients as they age, and that the above analyses
investigating the effects of delayed RP would need to be stratifi ed by
comorbidity status.
Using patient age, baseline, and postoperative IIEF-6 scores from the
study cohort, we estimated the expected postoperative erectile function
recovery using locally weighted scatterplot smoothing for immediate
surgery versus surgery after varying periods of observation. An average
IIEF-6 score over 10-yr duration was calculated for the two scenarios by
calculating the area under the erectile function curve. Bootstrap
resampling was used to 95% confidence intervals (CIs). Our plan was
to use prior data on the distribution of time to treatment to give
expected erectile function on active surveillance. In brief, the predicted
IIEF score over 10 yr for a patient undergoing surgery after nyr would be
multiplied by the actuarial probability of surgery at n yr with results then
summed over all n. However, this analysis was found not to bewarranted.
All statistical analyses were conducted using Stata 13 (Stata Corp.,
College Station, TX, USA).
E U R O P E A N U R O L O GY 7 3 ( 2 0 18 ) 3 3 – 3 7 34