Transcript Adjuvant Guidelines

These slides were released by the speaker
for internal use by Novartis
Tools for assessing risk of
relapse in individuals
Peter Ravdin
(University of Texas Health Science Center at San Antonio, TX, USA)
Adjuvant Guidelines
(Never A Mention Of Numbers)
A Relic Of The Empire !
Clin Pract Guide Oncol. v.1.2006. Breast Cancer. http://www.nccn.org;
Goldhirsch et al. Ann Oncol 2005;16:1569–83
What Is Missing From These Guidelines?
Quantitative numerical estimates of
benefit gained or given up….
Ravdin et al. Lancet 2002;359:2126–7.
Change In The Goals Of Prognostication
Effective Adjuvant Therapy!
But Is A Given Adjuvant Therapy Worth It?
Cost,
Toxicity
X % OS
Benefit
• Chemo: Leukemia 0.3% mortality, Sepsis 0.1%, CHF 0.1% ?
• Tamoxifen about a 0.2 % mortality (Thrombosis/Uterine CA)
% Women Satisfied With
This Amount
How Much Of A Reduction In Breast Cancer
Would Make The Adjuvant Worthwhile?
50
Bimodal Distribution
Of Answers
40
30
20
10
0
<0. 5
< 0.5
. 5 -1
0
- 2
1
1–2
2- 5
5- 1
0
0 - 20
1
5–10
>20
< 20
0.5–1.0
2–5
10–20
% Reduction Breast Cancer Mortality
Ravdin et al. J Clin Oncol 1998;16:515–21
First Widely Used Tool For Prognostic Estimates:
Nottingham Prognostic Index
0.2 * Tumor size in centimeters
+ Stage of lymph nodes (1 to 3 by level )
+ Histologic grade (SBR, 1-3)
______________________________________
Nottingham Prognostic Index
Group
Excellent
Good
Moderate
Poor
Score
< 2.4
< 3.4
3.41–5.4
> 5.4
15 yr BCSS
15 %
20 %
58 %
87 %
For NN Patients = Scores from 2.0–5.0
NN 2.0 cm Grade 2 = 3.4 (Good)
Galea et al. Breast Cancer Res Treat 1992;22:207
Tools For Prognostic Assessment
And Decision Making
Adjuvant!
Whelan Decision Boards
Nottingham Index
Finn Prog
MSKCC
Mayo Model
Adjuvant!
A program for aiding health professionals in making
estimates of outcome of patients with invasive cancer who
have undergone definitive local therapy (without prior
radiation or systemic therapy) and who are now deciding on
whether to get systemic adjuvant therapy
Ravdin et al. J Clin Oncol 2001;19:980
Mainscreen
Information Input
Natural Mortality
Tx Efficacy
Br Ca Mortality
Age and Average Non-Breast Cancer Mortality
(at 10 years Follow-up)
Mortality (%)
60
50
40
30
20
10
0
40
50
60
Age At Start
70
80
Risk Estimates In Adjuvant!
Derived From SEER for N0T1c Cases
Breast Cancer Deaths at 10 Years
SEER 2001
But It Has Not Included A
Variable That Is Important
What About Her2?
What About Tumor Detection Method?
In areas where there is controversy
you make the choice
With some assistance of the help files
Logically Using Additional
Prognostic Information
Her2 Prognostic Review in Help Files
Largest Study: From Slamon’s Lab
589 untreated node-negative patients
Vysis FISH used
Published JCO 2000 18:86–96
Her2 was a weak independent variable
Pauletti et al. J Clin Oncol 2000;18:3651
Using Her2
Mammographically Detected Tumors
Combined analysis of 3 randomized trials
1927 breast cancer cases – most without adjuvant
Shen et al. J Natl Cancer Inst 2005;97:1195–203
Detection method was a weak independent variable
Finnish non-randomized study found a RR of 1.9
Should State Evidence / Assumptions
Setting for Validation Study
BC Cancer Agency
• Population about 4 million
• 2600 new breast cancers/year
• 75–85% referred to BC Cancer Agency
• Breast Cancer Outcomes Unit
Systemic therapy indications
1989–92
•Node positive
•pN0 if LVI +
•pN0 if T>2cm and ER negative
•If age >65 years, not given chemo
Olivotto et al. J Clin Oncol 2005;23:2716
Results: Overall effect: N=4083
Adjuvant! BCOU
Predicted Observed Pred-Obs
10-yr OS
71.7%
72.0%
-0.3%
10-yr BCSS
83.2%
82.5%
+0.7%
10-yr EFS
71.0%
70.1%
+0.9%
All p = NS
Breast Cancer-Specific Survival
Slopes of a perfect fit line (red) and a line fitted to the
observed data (blue) were not different
100
BCOU
Observed BCSS
BCOU Observation
90
80
70
60
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
Adjuvant! Estimate
Adjuvant! predicted BCSS
90
100
10-year BCSS: Age
Number
Adjuvant!
BCOU
Pred - Obs
Predicted Observed
20-35 yrs
127
78.1%
68.5%
+9.6%*
36-50 yrs
1117
81.0%
81.0%
0%
51-65 yrs
1372
83.6%
84.1%
-0.5%
66-75 yrs
1070
84.5%
83.7%
+0.8%
76+ yrs
397
86.6%
82.6%
+4.0%*
Age <35: Adjuvant! was 1.5X optimistic
Age >75: Adjuvant! was 1.3X optimistic
*p<0.05
Weaknesses of Adjuvant!
Input Variable Issues
 Categorical use of T and N subgroups
 Use of histologic grade as a categorical
variable with “errors” at interfaces
 Possible drift in variables such as nodal
status miscalled in about 10% of NN
patients who have SLNB
These non-idealities affect both guideline and
tool-based decisions
Weaknesses of Adjuvant!
Limited knowledge about treatment regimens
 Mid range follow-up on new regimens?
 Equally effective in all patient subsets?
These non-idealities effect both guideline and
tool-based decisions
Making Efficacy Estimates
Selecting Options for Therapy
(Proportional Risk Reductions)
2000 Overview
Effectiveness Of Adjuvant Therapy
Tamoxifen Chemo Combined
< 50
ER+
ER-
32 %
0%
30 %
30 %
48 %
30 %
ER+
ER-
32 %
0%
10 %
18 %
39 %
18 %
> 50
Anonymous. Lancet 1998 351:1451–67
Anonymous: EBCTCG Lancet 1998 352:930–42
The Generations: Hormonal Therapy
(Lineages and Chains of Inference)
Tamoxifen
Ovarian Strategies
Aromatase
Inhibitor
Strategies
?
The Generations
(Lineages and Chains of Inference)
CMF
CMF
FE(50)C
CA * 4
CAF,
CEF
FAC
FE(100)C
CA*4+P*4
DAC
FEC*3+D3 CA*4+P*4 q2w
FEC*4+P*8 CA*4 +P*12qw)
P = paclitaxel; D = docetaxel; A = doxorubicin; E = epirubicin
The Generations
Trials Comparing Regimens
CMF
CMF
FE(50)C
CA * 4
CAF,
CEF
FAC
FE(100)C
CA*4+P*4
DAC
FEC*3+D3
Q2W (?)
(CA*4+P*4)
P = paclitaxel; D = docetaxel; A = doxorubicin; E = epirubicin
Selecting A Treatment
Its Flexible! States Assumptions/Data!
Chemotherapy Has Less Almost
No Late Effect In Older Women
50 +
36%
1 %*
8%
16 %*
EBCTCG. Lancet 2005;365:1687–717
What Probably Will Not Be Part Of The
2006 Overview But Still Is A Hot Topic
Adjuvant trastuzumab
Some US Clinicians State
“All Her2 Positive Patients Should Get
trastuzumab”
Is this reasonable and what does Adjuvant!
say about this??
Combined Analysis for OS of
NSABP B-31 / NCCTG – N9831
ACTH
94%
91%
ACT
92%
87%
ACT
ACTH
N
1679
1672
Deaths
92
62
HR=0.67, 2P=0.015
Years From Randomization
B31/N9831
Romond et al. N Engl J Med 2005;353:1673–84
Cardiac Monitoring
Age and Post AC LVEF were predictors
of the risk of developing CHF
Risk of CHF (%)
Age younger
than 50
Age 50 and
older
Initial LVEF 50 - 54
6.3 %
19.1 %
Initial LVEF 55 - 64
2.2 %
5.2 % *
Initial LVEF > 65
0.6 %
1.3 *
In both age groups about 10% of the patients had a LVEF of 50-54,
about 50% of the patients had a LVEF of 55-64, and 35% had a LVEF
of > 65%. Average risk of early CHF for patient younger than 50 is
2% and older than 50 is ~ 5%
So Is Adjuvant Herceptin For All Breast
Cancer Patients? Informed Speculation !
60 Year Old Women: ER +, Her2 +, average comorbidity
Competeing mortality about 8%: To Get Tam + CA * 4, T * 4q3w
Her2 FISH +: Additional RR conferred by Her2 1.5
Baseline 10
Year Risk of
Death
With Tam
and Chemo
Added
trastuzumab
Benefit Due
to
trastuzumab
NN T1c
19 % (11%)
14 % (6%)
12 % (4%)
2%
NN T1ab
12 % (4%)
10 % (2%)
9 % (1%)
1%
Risk of developing CHF ~5%, 2/3 have symptoms resolve
in 6 months. Cardiac status at 10 years??
The Crucial Question Is Not Which
Regimen Is Best…
The Real Question Is, Can We Tell
Which Patient Would Most Benefit
From Which Regimen?
Do ER Level / Her2 Expression / Specific
Genomic Profiles Predict Responsiveness
Emerging Picture Of Breast Cancer Subtypes
And Treatment Efficacy
( St Gallen Guidelines )
ER -
Endocrine
Non-Responsive
ER +
Endocrine
Response
Uncertain
Endocrine
Responsive
St Gallen Guidelines
Definition Of Endocrine Responsiveness Uncertain
An Interesting Mixed Bag Of Features
Low ER
No PgR
Her2 + (for Tamoxifen)
Large Number of Nodes
The exact boundary between “endocrine responsive” and
“ endocrine response uncertain” is unknown
What Is New: Genomic Profiles
An example of what should be better.
Oncotype Dx
Excellent Standardization
Multiple quantitative measured variables
Continuous rather then categorical
Uses well defined data sets
Ravdin 2005
Adjuvant! Genomic Variant
Print Schemata and Side-Effects Information
Conclusions
Decision tools have powerful advantages
over guidelines
Decision making depends on integrating
increasingly complex information about:
Prognosis, treatment efficacy, toxicity
and competing mortality
And communicated this information in an
intelligible manner