|
Percent Of Biopsy Cores That Are Positive Improves Prediction (September
2003)
There is general
agreement that when choosing data on which to base predictions of
pathology and treatment outcome the percent of biopsy cores that are
positive is more informative than the clinical stage. The percent positive
figure can replace clinical stage in the predictive trio that includes
Gleason sum and PSA. It's intuitive that clinical stage T1c hides a wide
spectrum of pathology. As we move further into the PSA era, it is not
surprising that the clinical stage is becoming less of an independent
predictor of pathologic stage since the majority of men are currently
presenting with clinical stage T1c at the time of diagnosis and earlier
tables are based on older data. "There is less spread in clinical stage
categories making the traditional Partin table less accurate" (Gancarczyk,
UROLOGY, March 2003). It had been rumored that Michael Kattan, PhD, was
going to augment the new Partin tables by adding "percent positive" data,
but in personal communication with Dr. Kattan, I was referred to his
article "Prediction of Progression: Nomograms of Clinical Utility
(Clinical Prostate Cancer, September, 2002) where Kattan and Scardino
conclude "We recently examined the impact of adding percent of cores
positive to our preoperative nomogram. We found this variable to be
statistically significant (P < 0.001), knocking out clinical stage, which
became statistically insignificant (P > 0.05). However, the concordance
index for this new monogram was identical to the original nomogram,
suggesting no improvement in overall predictive accuracy after inclusion
of percent of positive cores."
Two consortiums,
however, have introduced "percent positive" figures into their predictive
models. The SEARCH Database data (J UROL, June, 2003) based on 1094 men
treated in four California centers stratified their data into categories
wherein <34%, 34%-50%, and >50% of cores were positive, and for them
"percent positive" data was an independent predictor of PSA recurrence
post prostatectomy. The second article (Gancarczyk, ibid) presented the
data from The Center for Prostate Disease Research (CPDR) based on the
correlation of 1510 prostatectomy specimens and preoperative biopsy
findings and incorporated their results in a nomogram predicting
pathologic stage after RP. Their nomogram stratifies data into groups with
>30%, 30%-59%, and >60% positive biopsies and their results differ from
the Partin table predictions in quite a few categories.
It would seem that
this "percent positive" data would be most relevant to clinical decision
making in instances where patients with seemingly low risk (i.e. T1c) may
present with a spectrum ranging from one or two cores positive compared to
(say) all cores positive. Conversely, for those men with features of high
risk disease, where the probability of involvement of nodes and seminial
vesicles might influence the choice of management, it may be useful to
consult the CPDR tables. The Gancarczyk article offers an example from the
low risk category: A 62 year old man, PSA 4.1 ng/ml, biopsy Gleason sum 6,
with 6 of 10 cores positive: probability of organ confined disease - 52%;
capsular invasion - 39%, +SV - 5%, and +N - 5%. Comparable parameters
entered into the new Partin tables (T1c, Gleason 6, PSA 4.1 ng/ml)
indicate probabilities of OC 80%, Cap+ 19%, SV+ 1%, N+ 0%. An example in
the higher risk category: CPDR data - PSA 15, Gleason sum 8, 3 of 12 cores
positive: OC 33%, Cap+ 35%, SV+ 25%, N+ 8%. The new Partin data (UROL,
December, 2001), which uses clinical stage instead of "percent positive"
cores predicts for PSA 15, Gleason sum 8, T2b (TNM'92): OC 7%, Cap+ 46%,
SV+ 19%, N+ 27%.
BOTTOM LINE:
No predictive model can be expected to be perfectly accurate. Now that the
number of biopsy cores that are positive is routinely included in
pathology reports, clinicians might want to consult both the Partin
and CPDR tables when making disease management decisions.
«
Back to Article List |
