1. RESULTS: The potential reduction in unnecessary biopsies for
the entire population using this approach was 80% based on deceler-
ation and 72% based on decrease below the level prior to biopsy. The
reduction varied substantially among the groups of exponential growth
rates in PSA. For example, reductions for decrease started at 59% for
very slow exponential PSA growth rates of 0%-5% and reached 93% for
fast PSA growth greater than 100% per year.
CONCLUSIONS: This analysis suggests that a delay in a bi-
opsy to allow additional PSA testing provides useful information about
PSA trends. In many cases, subsequent PSA trends showed a sub-
stantial deceleration or decrease that could have reduced the number of
unnecessary biopsies by 72-80%. The fastest growth in PSA per year
was also the most likely to decrease, with potential reductions of 93%
in biopsies.
Source of Funding: NONE
MP63-04
NEW ANALYSIS OF PSA TRENDS HELPS IDENTIFY DEADLY
CANCERS PRIOR TO BIOPSY
Lori Rawson*, Roy MacKintosh, Reno, NV; Christopher Morrell,
Baltimore, MD; R. Jeffrey Karnes, Rochester, MN; Stacy Loeb,
New York, NY; Stephen Van Den Eeden, Oakland, CA; Thomas Neville,
Incline Village, NV
INTRODUCTION AND OBJECTIVES: A major prostate cancer
screening challenge is to balance the potential harm of biopsies and
over-treatment with the need to identify deadly cancers early for
effective treatment. Observation of men with deadly cancer and their
prior PSA levels often appears to show exponential growth in PSA
over and above baseline PSA of a benign nature. Preliminary
research on small populations has suggested that faster exponential
growth rates in PSA above a no-cancer baseline are indicative of
more deadly cancers. Our objective was to validate the relationship
between dynamic PSA trends and later prostate cancer mortality on a
large population.
METHODS: We analyzed Veterans Affairs data on 58,523 men
age 50-75 (median 66) who had been diagnosed with prostate cancer
between 2001-2012 with at least three PSA tests over at least two years
prior to diagnosis. We fit an exponential plus constant trend to the PSA
tests for each man. This process allowed us to estimate for each man
the amount of PSA that might be coming from cancer and its annual
exponential growth rate. Men were grouped by ranges of age, cancer
PSA and growth rates. For each group, we utilized Kaplan Meier
methods to estimate all-cause death risk for years after diagnosis.
Cancer-specific death was estimated for each group using net-survival
methods based on an estimate of cancer-free survival and were
compared by the log rank test.
RESULTS: For a given PSA range, the faster the growth in PSA
measured in this way, the more deadly the cancer. For every range of
cancer PSA, all-cause and cancer-specific death increased substan-
tially for faster PSA growth (p<0.0001). For example, Figure 1 shows
the robust relationship between dynamic PSA trends with later prostate
cancer death in >30,000 men ages 50 to 75. These results held up for
various age ranges and types of treatment.
CONCLUSIONS: Higher annual exponential growth rates in
cancer PSA were significantly associated with all-cause and cancer-
specific death in our large population-based cohort. For men with a PSA
history, the growth rate calculated using these methods is highly
indicative of which men harbor life-threatening prostate cancer. This
result may help address the challenge issued by the U.S. Preventive
Services Task Force to develop screening methods to distinguish
potentially deadly cancers from non-progressive or slowly progres-
sive disease.
Source of Funding: NONE
MP63-05
SERUM TESTOSTERONE IMPROVES THE ACCURACY OF
PROSTATE HEALTH INDEX FOR THE DETECTION OF PROSTATE
CANCER
Frank Friedersdorff*, Kurt Miller, Klaus Jung, Carsten Stephan, Berlin,
Germany
INTRODUCTION AND OBJECTIVES: The detection of pros-
tate cancer (PCa) hampers on low specificity when using prostate-
specific antigen (PSA) alone. This study investigated the effects of total
testosterone (tT) and its free (fT) and bioavailable (bioT) subforms in
serum to improve the diagnostic validity of serum (-2)pro-PSA-based
prostate health index (PHI).
METHODS: Total and free PSA (tPSA, fPSA), (-2)pro-PSA,
testosterone, and sex-hormone-binding protein were measured by
automated immunoassays from the serum of 193 men scheduled for
prostate biopsy (99 with PCa and 94 with no evidence of malignancy).
FT and bioT were calculated using an online calculator. Statistical an-
alyses were performed by non-parametric tests (Wilcoxon signed rank,
Mann-Whitney, Kruskal-Wallis), binary logistic regression, and receiver
operating characteristics (ROC) analyses.
RESULTS: Compared to the non-malignant controls, PCa pa-
tients had significantly higher levels of tPSA and PHI, but lower levels of
%fPSA, tT, fT, and bioT. PCa could be differentiated from controls by
PHI, tT, fT, bioT, and %fPSA. PHI showed the largest area under the
ROC curve (AUC ¼ 0.73) that was additionally increased by the in-
clusion of bioT or tT in a binary logistic regression model. The AUC of
PHI in patients with tT concentrations of <8 nmol/L indicating
biochemical hypogonadism was significantly larger than that in patients
with higher tT values (0.86 vs 0.70).
CONCLUSIONS: The PHI based discrimination between PCa
patients and non-malignant controls could be improved by the simul-
taneous determination of testosterone, while patients with testosterone
concentration of <8 nmol/L have the greatest advantage.
Source of Funding: none
e710 THE JOURNAL OF UROLOGYâ Vol. 191, No. 4S, Supplement, Monday, May 19, 2014