PSA Screening and The Changing Role of Surgery for Prostate Cancer Peter T. Scardino, MD
Chairman, Department of Surgery and Head of the Prostate Cancer Program Memorial Sloan-‐KeAering Cancer Center New York, NY
Disclosure
I am a scienGfic adviser to OPKO, a biomedical company that licensed the 4-‐kallikrein panel of PSA-‐related biomarkers (PSA, free PSA, intact PSA and hK2) developed at Memorial Sloan-‐ KeAering Cancer Center to improve the accuracy of PSA tesGng.
Screening RecommendaGons before 2010
The ACS, AUA and NCCN recommend that men should first be informed of the risks and potenGal benefits of screening, then offered PSA and DRE beginning at age 50 years and conGnuing annually as long as they have a life expectancy > 10 years. The discussion should take place at age 45 for men at high risk (e.g., African-‐Americans and men with a family history), and at age 40 for very high-‐risk men (e.g., several first-‐degree relaGves).
Cancer incidence and mortality rates Age-‐adjusted, standardized over Gme
hAp://www.cancer.org/Research/CancerFactsFigures/cancer-‐facts-‐figures-‐2012
U. S. PrevenGve Services Task Force (USPSTF) Screening for Prostate Cancer Grade: DefiniGon:
D The USPSTF recommends against screening for prostate cancer with PSA. There is moderate or high certainty that there is no net benefit or that the harms outweigh the benefits.
SuggesGons for pracGce: Discourage the use of PSA screening.
What was wrong with PSA screening the way it was pracGced? Screening and consequent treatment have oaen been harmful: • • • •
Too much screening of elderly men with a short life expectancy Too liberal criteria for biopsy Too aggressive treatment of low-‐risk cancer Treatment largely administered by low-‐volume providers (higher risk of side effects and lower chance of cure)
Vickers AJ, Lilja H. Time for another rethink on prostate cancer screening. Nat Rev Clin Oncol 2011;9:7-‐8. Carlsson S, Vickers AJ, Roobol M et al. Prostate cancer screening. J Clin Oncol 2012;30:2581-‐4.
EsGmated prevalence of PSA screening by year and age
Drazer MW, Huo D, Schonberg MA, et al. PopulaGon-‐based paAerns and predictors of PSA screening among older men in the United States. J Clin Oncol 2011; 29: 1736-‐1743.
PSA variability (annual test)
Percent of men on a trial of dietary intervenGon for colon polyps with abnormal PSA during 5-‐year period (n=972) and probability that PSA would later return to normal and remain normal
Defini'on of abnormal PSA
% ever abnormal
Probability that PSA would
Probability that PSA would
Return to normal
Remain normal
% fPSA
> 4.0 ng/mL > 2.5 ng/mL Age specific (PSA 4-‐10)
Any
21% 37% 20% 15%
44% 40% 55% 53%
37%
Modified from Eastham JA et al. JAMA 2003;289:2695.
80% 65% 83% 74%
Epidemiologic esGmates of overdetecGon
J Natl Cancer Inst 2009;101:374-383
Of all screen-‐detected cancers, an esGmated 10-‐56% would not have become apparent or caused symptoms in the paGents’ lifeGmes, and the lead Gme for diagnosis is esGmated at 5-‐12 years.
ComplicaGons following prostate biopsy 30-‐day
Hospitaliza'on rate
Infec'on
Bleeding
Urinary obstruc'on
5% random sample of Medicare pts, 1991-‐2007 (n= 17,472)
0.38
Not reported
Not reported
All men having prostate bx not resul'ng in Ca dx Ontario 1996-‐2005 (n= 41,682)
Study popula'on
Consecu've pts at single center 2001-‐10 (n=1000)
6.9
1.9
2.5
Loeb S et al. J Urol 2011;186:1830-‐1834. Nam RK et al. J Urol 2010;183:963-‐968. Pinkhasov GI et al. BJU Int 2012;110:369-‐374.
1.36
1.2
0.37
0.4
0.17
0.8
New Engl J Med 2008;358:1250-‐61
• Sexual funcGon caused moderate or severe distress aaer 2 years in 43% of men aaer RP, 37% aaer XRT and 30% aaer brachy; ED caused distress in 44%, 22% and 13% of partners
• Urinary funcGon led to moderate or severe distress aaer 1 yr in 7% of paGents aaer RP, 11% aaer XRT and 18% aaer brachy and in 4-‐5% of partners
• GI symptoms caused moderate or severe distress in 9% of paGents (and 4-‐5% of partners) 1 year aaer radiaGon (XRT or brachy)
Use of watchful waiGng among U.S paGents with localized prostate cancer by risk category (1990-‐2006)
Low
Intermediate
High
Cooperberg MR, Broering JM, Carroll PR. Time trends and local variaGon in primary treatment of localized prostate cancer. J Clin Oncol 2010; 28: 11117-‐23.
What is wrong with the USPSTF recommendaGons? • European randomized trial had not reached its primary endpoint • Inadequate consideraGon of Gme-‐to-‐event (long delay between elevated PSA and death from prostate cancer) • Lack of appreciaGon for the strong relaGonship between PSA and prostate cancer mortality • Overall mortality is not an appropriate endpoint • Mortality risk of surgery exaggerated (“0.5%”) Vickers AJ, Lilja H. Nat Rev Clin Oncol 2011;9:7-‐8
Does PSA screening reduce prostate cancer mortality? PLCO: No
Does PSA screening reduce prostate cancer mortality? ERSPC: Yes, but at high cost
Over a median of 9 years, PSA screening reduced deaths from prostate cancer by 20% but was associated with a high risk of over-‐ diagnosis: 1410 men needed to be screened and 48 cases diagnosed or treated to prevent 1 death from prostate cancer.
Over 14 years, PSA tesGng increased the number of cancers detected by 50% but reduced the risk of dying of prostate cancer by 44% in all men and 56% in screened men. The number needed to screen to prevent one death was 293, and the number needed to diagnose or treat was 12.
Lancet Oncol 2011
RESULTS
Per 1000 men of all ages who were followed for their entire life span, we predicted that annual screening of men between the ages of 55 and 69 years would result in nine fewer deaths from prostate cancer (28% reduction), 14 fewer men receiving palliative therapy (35% reduction), and a total of 73 life-years gained (average, 8.4 years per prostate-cancer death avoided). The number of QALYs that were gained was 56 (range, -21 to 97), a reduction of 23% from unadjusted life-years gained. To prevent one prostate-cancer death, 98 men would need to be screened and 5 cancers would need to be detected. Screening of all men between the ages of 55 and 74 would result in more life-years gained (82) but the same numbers of QALYs (56).
Screening trials
Number needed to screen or diagnose to prevent one death
Study
RR (risk ra'o)
___
PLCO
1.01
Screen _________
Diagnose or Treat a
Dura'on (years) 9
1410 565
EurUrol 2013
.80 .80b
48 33
9 13
ERSPC NEJM 2009 Göteborg LifeGme model (age 55-‐69)
.56c .72
293 98
12 5
14 lifelong
a 30% -‐ 40% placed on acGve surveillance b RR .68 in 55-‐ to 69-‐y.o. paGents
C 0.44 for men who were actually screened
Andriole GL et al. NEJM 2009; Roobol MJ et al. NEJM 2009; Roobol MJ et al. Eur Urol 2013; Hugosson J et al. Lancet Oncol 2010; Heijnsdijk EAM et al. NEJM 2012.
Probability of eventually developing or dying of prostate cancer by PSA at age 60
Mid-‐life PSA levels strongly predict long-‐term risk of prostate cancer morbidity
Clinical cancer AUC 0.74
ProporGon of populaGon with PSA level within each range; median is 1.0
Prostate cancer death AUC 0.89
at age 60 Vickers A, Lilja H et al. BMJ 2010;341:c4521.
PSA at age 45-‐49 strongly predicts risk of metastasis within 25 years Top 10%
Top 25%
Second 25% Third quarGle BoAom quarGle
Vickers AJ, Lilja H et al. Strategy for detecGon of prostate cancer based on relaGon between prostate-‐ specific anGgen at age 40-‐55 and long-‐term risk of metastasis: case-‐control study. BMJ 2013;346:2023.
How to increase the benefits and reduce the risks of screening for prostate cancer • Risk-‐adjust screening by age and PSA (ignore PSAV) (Reduce false posiGves)
• Add addiGonal markers to increase specificity: 4-‐kallikrein panel, phi, or urinary PCA3 (Reduce indicaGons for biopsy)
• AcGve surveillance for low-‐risk cancers (Reduce harms of unnecessary surgery)
• Treatment by high-‐volume physicians and centers (Reduce harms of necessary therapy) Vickers A, Roobol M, Lilja H. Ann Rev Med 2012; 63:161-‐70.
Memorial Sloan-‐KeAering Cancer Center 2010 Guidelines Risk-‐adjusted screening for prostate cancer
Begin PSA tesGng at age 45
For men age 45-‐59
• PSA ≥ 3 ng/mL : consider biopsy • PSA > 1 but < 3 ng/mL : return for PSA every 2-‐4 years • PSA < 1 ng/mL : return for PSA in 5 years or at age 50 or 60, whichever comes first
For men age 60-‐70
• PSA ≥ 3 ng/mL : consider biopsy • PSA > 1 but < 3 ng/mL : return for PSA every 2 years • PSA < 1 ng/mL : no further screening
For men age 71 or more
• No further screening unless previously unscreened
PSA markers Standard markers
tPSA = total PSA, the standard measure
%fPSA = raGo of free to total PSA
New marker panels
4K score = tPSA, fPSA, iPSA, hK2*
phi = ([-‐2] proPSA/fPSA x √tPSA)** *OPKO DiagnosGcs, Inc. ** Beckman Coulter, Inc.
The 4K score helps to reduce the number of biopsies while missing few high-‐grade, clinically significant cancers 4K score* = tPSA, fPSA, iPSA, hK2
Biopsies Biopsy all if PSA ≥3 Biopsy only if risk of cancer >20% on 4K score
Cancers
High-‐grade cancers
Total
Reduced
Total
Missed
Total
Missed
740
-‐
192
-‐
40
-‐
316
424 (57%)
161
31 (16%)
37
3 (8%)
Results based on laboratory model only, not including DRE *OPKO DiagnosGcs, Inc. Vickers AJ et al. BMC Med 2008;6:19.
ValidaGon of 4K score for predicGng biopsy outcomes 7 studies with 10,575 biopsies and 2860 cancers
AUC
AUC* (PSA + DRE)
AUC (4K+DRE)
(high-‐grade)
(RoAerdam) (Goteborg)
0.70 0.72 0.59 0.62
0.78 0.84 0.71 0.70
0.84 0.90 0.80 0.83
Prior negaGve biopsy (RoAerdam)
0.58
0.68
0.87
(ERSPC Tarn)
0.63
0.78
0.87
Unscreened men, clinical PCa diagnosed within 5 years (Sweden)
0.65
0.75
0.82**
Cohort Unscreened
(RoAerdam, NL) (Goteborg, Sweden)
With prior screening
Clinical work up before biopsy
* AUC: area under the curve, or accuracy ** Advanced prostate cancer, i.e., clinical stage ≥ T3 or evidence of metastases at Gme of diagnosis Vickers AJ et al. Cancer Epidemiol Biomarkers Prev 2011;20:255-‐61.
(4K+DRE)
New PSA markers:
[-‐2] Pro-‐PSA panel (Prostate Health Index: phi) A mulGcenter study of [-‐2] Pro-‐PSA combined with PSA and free PSA for prostate cancer detecGon in the 2.0 to 10.0 ng/mL PSA range Catalona WJ et al. J Urol 2011;185:1650
In the 2 to 10 ng/mL PSA range at 80% to 95% sensiGvity, the specificity and AUC (0.703) of phi exceeded those of PSA and %fPSA.
Actuarial esGmate of remaining on AS
mulG-‐insGtuGonal retrospecGve study of candidates with very low risk cancer 1.00
100%
0%
0.75
0.50
Five-‐year: 75%
0.25
25%
Two-‐year: 91%
0.00
50%
Free from acGve treatment
Proportion of men
75%
Actuarial Estimate of Remaining on Active Surveillance
At risk: 261 0
Eggener S et al. J Urol 2009
158
81
20 40 Time since active surveillance (months)
39
22
60
80
Time to BCR for treated paGents who progressed on AS
50.4% of 117 treated paGents and 13% of the overall cohort had PSA recurrence aaer delayed therapy
All-‐cause mortality for men with conservaGvely managed prostate cancer Charlson comorbidity index 0
Rider JR et al. Long-‐term outcomes among non-‐curaGvely treated men according to prostate cancer risk category in a naGonwide populaGon-‐based study. Eur Urol 2013;63:88-‐96.
PIVOT: RP v. ObservaGon Overall results: All-‐cause and prostate cancer-‐specific mortality
N = 731, mean age 67
Wilt TJ et al. N Engl J Med 2012;367:204-‐13.
Trends in iniGal treatment of low-‐risk prostate cancer
Memorial Sloan-‐KeAering
Silberstein Jl et al. Reverse stage shia at a terGary care center: escalaGng risk in men undergoing radical prostatectomy. Cancer 2011;117:4855-‐60.
United Kingdom
McVey GP et al. Temporal trends in iniGal treatment of low-‐risk prostate cancer. BJU Int 2010;106:1161-‐64.
Smarter Screening with PSA We can screen more effecGvely, keeping the mortality rate from prostate cancer low without harming large numbers of men, by screening smarter: • Risk-‐adjusGng the frequency of tesGng and repeaGng an elevated test aaer 6-‐12 weeks before recommending biopsy (ignore PSAV) • Using biomarkers with greater specificity to reduce false-‐posiGve tests and unnecessary biopsies • Expanding acGve surveillance for low-‐risk cancers and in older men • Referring men with aggressive cancers to regional centers for high-‐quality care
The changing role of surgery • RP should rarely be used for low-‐risk, organ-‐confined cancers
Restage with MRI and repeat biopsy to rule out an aggressive cancer, then monitor with an AS protocol.
• Properly performed, RP is a highly effecGve treatment for intermediate-‐ and high-‐risk cancer in men with a life expectancy >10 yrs
Randomized trial of RP v. WW (2014)
Radical prostatectomy significantly reduced the risk of death from any cause (HR 0.71), death from prostate cancer (HR 0.56) and metastases (HR 0.57) compared with WW
RP also dramaGcally reduced the risk of clinical local progression (HR 0.34) and the need for hormone therapy (HR 0.49).
The absolute risk reducGon for death from any cause was 12.7%, death from prostate cancer 11.0%, distant metastases 12.2%, clinical local recurrence 27.9% and hormone therapy 25% Bill-‐Axelson A et al. Radical prostatectomy versus watchful waiGng in early prostate cancer. New Engl J Med 2011;364:1708-‐17; Bill-‐Axelson A et al. Radical prostatectomy or watchful waiGng in early prostate cancer. New Engl J Med 2014;370:932-‐42.
Number needed to treat (NNT) with RP to prevent one death Dura'on (years) All ages
< 65 years old
5
50
10
20
8
15
15
7
18
8
4
Holmberg L et al. NEJM 2002; Bill-‐Axelson A et al. New Engl J Med 2005; Bill-‐Axelson et al. New Engl J Med 2011; Bill-‐Axelson A et al. New Engl J Med 2014.
PIVOT: RP v. ObservaGon
Intermediate-‐ and high-‐risk groups Intermediate-‐risk
High-‐risk HR 0.78 P = .039
HR 0.89 P = .16
All-‐cause mortality
Prostate cancer-‐ specific mortality
Metastasis:
HR 0.57 P = .12
Intermediate-‐risk HR 0.56; P = 0.08
HR 0.52 P = .08
High-‐risk HR 0.36; P = 0.01
Supplement to: Wilt TJ, Brawer MK, Jones KM, et al. Radical prostatectomy versus observaGon for localized prostate cancer. N Engl J Med 2012;367:203-‐13.
Model of cumulaGve incidence of prostate cancer death for watchful waiGng and RP groups (A), in the absence of PSA screening (as in SPCG-‐4) and (B), with screening (as in PIVOT).
Long-‐term cancer control and survival aaer RP for “high-‐risk” cancer Yossepowitch et al. J Urol 2007; 178: 493.
Freedom from BCR
Prostate cancer-‐specific survival
Incidence of +LN with full PLND in a contemporary RP series (MSK) Number (%) of pa'ents
Number (%) with +LN
3527
194 (5.2%)
Overall AUA risk group
Number (%) of pa'ents
Number (%) with +LN
% of +LN
Low risk
1693 (48%)
8 (0.5%)
4%
Intermediate risk
1478 (42%)
72 (5%)
37%
High risk
550 (16%)
114 (21%)
59%
Limited v. full PLND:
Removal of a few more nodes substanGally increases the % of paGents with +LN Limited PLND (PLNI > 2%)
Standard PLND (PLNI > 2%)
9 IQR 6-‐13
13 IQR 9-‐17
5% (95% CI: 1%-‐8.9%).
13.2% (95% CI: 8.7%-‐17.6%)
Median number of nodes retrieved Rate of LN metastasis
Ext il
H
Touijer KA et al. J Urol 2006
P value
p= 0.016
Long-‐term outcome aaer RP + PLND alone for paGents with lymph node metastasis
PFP at 10 yrs: 28% (95% CI, 21%–36%)
Touijer KA et al. Long-‐term outcomes of paGents with lymph node metastasis treated with radical prostatectomy without adjuvant androgen-‐deprivaGon therapy. Eur Urol 2014;65:20-‐25
Key prognosGc factors for paGents with +LN: RP Gleason score and number of +LN
von Bodman C et al. PredicGng biochemical recurrence-‐free survival for paGents with posiGve pelvic lymph nodes at radical prostatectomy. J Urol 2010;184:143-‐8.
Analysis of prostate cancer-‐specific mortality aaer RP and PLND
.2 Death Rate .6 .4
.6 .4
1
0
.8
.2
Survival
.8
0
1
Age 60-69, Lymph Node Positive
0
5
10 Years After Surgery PC Death Survival
15
20
Non-PC Death
Death rate of prostate cancer and other causes at 10, 15 and 20 yrs aaer surgery with 95% CI Prostate cancer Other causes
10 year 0.12( 0.077, 0.17) 0.072 (0.040, 0.12)
15 Year 0.22 (0.15, 0.30) 0.13 (0.079, 0.20)
20 Year 0.42 (0.26, 0.57) 0.16 (0.089, 0.26)
Eggener S et al. PredicGng 15-‐year prostate cancer specific mortality aaer radical prostatectomy. J Urol 2011;185:869-‐75.
Surgery for Prostate Cancer RP with PLND works well for intermediate and high risk, clinically localized prostate cancer. Cancer control is excellent, and long-‐term cancer-‐ specific survival is very good with early salvage radiaGon for PSA recurrence. But complicaGon rates are relaGvely high and side effects troublesome to paGents and their partners.
What determines outcomes aaer RP? RP is a technically challenging operaGon. To what degree are the results of RP – perioperaGve complicaGons, long-‐term urinary inconGnence and strictures, and erecGle dysfuncGon as well as surgical margins and long-‐term cancer control – related to surgical technique in addiGon to the characterisGcs of the paGent and the cancer. Does the technology maAer (open, lap, roboGc)?
The chances of curing a prostate cancer aaer surgery increase as the surgeon gains experience
Vickers A, Bianco F, Serio A, Eastham J, Schrag D, Klein E, Reuter V, KaAan M, Pontes J, and Scardino P: The surgical learning curve for prostate cancer control aaer radical prostatectomy. JNCI 2007; 99:1171
RR
((RP cases)
Variability in 5-‐yr progression-‐free probabiliGes among experienced surgeons (>40 cases) adjusted for case mix
p