DIETARY FAT REDUCTION INPOSTMENOPAUSAL WOMEN WITH PRIMARY
BREAST CANCER: PHASE III WOMEN’SINTERVENTION NUTRITION STUDY (WINS)
R.T. Chlebowski, G.L. Blackburn, R.E. Elashoff, C. Thompson, M.T.Goodman, A. Shapiro, A.E. Giuliano, N. Karanja, M.K. Hoy, D.W. Nixon.
For the WINS Investigators
March 15, 2006
Dietary Fat and Breast Cancer Risk
Carroll Cancer Res 1975;35:3374-3383.
Dietary Fat and Breast Cancer Risk
WINS: BACKGROUND
Preclinical and observational studies suggest dietary fatintake may be related to breast cancer outcome (1982) 1
Feasibility of achieving dietary fat reduction in breastcancer patients demonstrated in randomized trials(1987, 1993) 2, 3
A randomized phase III clinical trial evaluating dietaryfat reduction influence on relapse-free survival in earlystage breast cancer patients receiving standard cancercare was initiated in 1994.
1 Wynder, Cohen Nutr Cancer 3:195-199, 19822 Chlebowski, Nixon, Blackburn, et al Breast Cancer Res Treat 10:21-29, 19873 Chlebowski, Blackburn, Buzzard et al J Clin Oncol 11:2072-2080, 1993
WINS: STUDY HYPOTHESIS
An intensive dietary intervention targetingfat intake reduction will prolong relapse-free survival as compared to a controlcondition with minimal dietary counselingin women receiving conventional therapyfor localized and resected breast cancer
WINS: TRIAL DESIGN
Women 48-79 yrs
Early breast cancer
Primary surgery +/- RTx
Systemic therapy*
Dietary fat intake> 20% of calories
Recruitment 1994-2001, Median follow-up 60 months
* Tamoxifen required, chemoRx optional for ER+; chemoRx required for ER-.Strata: Nodal status; Systemic Rx; Sentinel node Chlebowski, Blackburn, Elashoff, et al Proc Amer Soc Clin Oncol 24:10,2005
Dietary Intervention(n=975) to reducefat intakemaintainingnutritionaladequacy
Control (n=1462)
Randomization60:40 within 365days from 1o
surgery n=2437
WINS: STUDY DESIGN
Phase III, randomized prospective multi-centerclinical trial funded as peer-reviewed RO1.
Primary Endpoint: Relapse-free survival
Defined as time from randomization to:
breast cancer recurrence- local-regional- distant- ipsilateral breast
new contralateral breast cancer
Secondary Endpoint: Overall survival
WINS: INCLUSION CRITERIA
Women 48 to 79 years of age
Histologically confirmed, resected,unilateral invasive breast cancer
Lymph nodes examined
Acceptable adjuvant systemic therapy
Dietary fat intake > 20% of calories
Able to accept either randomization
Provide informed consent
WINS: EXCLUSION CRITERIA
Tumor < 1 cm with negative nodes
Tumor > 5 cm with positive nodes
> 10 nodes positive
Pre-operative chemotherapy
Previous neoplasm except skin cancer orin situ carcinoma of cervix
> 365 days from primary surgery
WINS: SYSTEMIC THERAPY
ER Positive (by local laboratory results):- Tamoxifen 20 mg/d x 5 yrs required- Approved chemotherapy optional
ER Negative- Approved chemotherapy required
Approved chemotherapy regimens: - (CMF, AC, CAF/FAC, AC → T)
WINS: DIETARY INTERVENTION
Goal: To reduce dietary fat intakemaintaining nutritional adequacy(target 15% calories from fat)
Weight loss not an interventiongoal; not counseled on weightreduction issues
WINS: DIETARY ASSESSEMENT
Unannounced dietary 24-hour telephonerecalls by trained interviewers 1, 2
Data entered in Nutrition Data System(NDS ®) software by NutritionCoordinating Center 1
3 recalls collected over two weeks foreligibility then two recalls annually whileon study
1 Buzzard, Faucett, Jeffery, et al J Amer Dietetic Assoc 96:574-9, 19962 Winters, Mitchell, Smicklers-Wright, et al J Amer Dietetic Assoc 104:551-9, 2004
WINS: DIETARY INTERVENTION
Diet Group: women given a fat gram goal bycentrally trained, registered dieticiansimplementing a low fat eating plan 1, 2
Eight bi-weekly individual counseling sessionsand subsequent contacts every 3 monthsMonthly group sessionsSelf-monitoring of fat gram intake
Control Group: women had dietician contactsevery three months
1 Chlebowski, Rose, Buzzard, et al Breast Cancer Res Treat 20:73-84, 19922 Winters, Mitchell, Smiciklas-Wright, et al J Am Diet Assoc, 104:551-9, 2004
WINS: STATISTICALCONSIDERATIONS
Sample size: calculation (n=2502) based on assumed7.5% ↑ in RFS with 0.10 drop-in and 0.30 drop-out, for84% power at 2-sided alpha of 0.05
Unbalanced randomization (60% control, 40% diet) toenhance resource allocation
Planned interim analyses on 2/2000; 10/2000; 11/2001;10/2002; and 11/2003
Primary efficacy analysis was a log rank testsupplemented by a Cox proportional hazards model
WINS: STRATIFICATION
Lymph node status negative positive
Systemic adjuvant therapy tamoxifen alone tamoxifen plus chemotherapy chemotherapy alone
Sentinel node evaluation yes no
WINS: STUDY POPULATION
2437 women from 37 U.S. clinicalsites, randomized 60:40 975 to dietary intervention group 1462 to control group
34 ineligible (most based on timefrom Dx, size, margins) but allare included in analyses as intent-to-treat
ENDPOINTS IN ADJUVANT HORMONETRIALS
XXXXXXNSABPDFS
XXXXXXWINSDFS
XXXXXXATACRFS*
OXXXXXIESDFS
OOXXXXMA-17DFS
OOXXXXWINSRFS
Newprimarycancer**
Deathwithoutrecurrence
Contrabreast
Ipsilateralbreast
DistantLocal-regional
TrialEndpoint
RFS – Relapse-free survival; DFS – Disease-free survival; * Recurrence – free survival** Other than breast cancer;
Goss et al N Engl J Med 349, 2003; Coombes et al N Engl J Med 350, 2004; ATAC Lancet 2003;Fisher et al J Natl Cancer Inst 93, 2001
WINS: STUDY TERMINATION
Following grant cycle completion, results from 5th interim analysiswere presented to an agency review panel for potential continuationof trial as recommended by the WINS External Advisory Committee(EAC)
Funding was not obtained and clinical center funding ended in May2004
The WINS External Advisory Committee reviewed results and statusof trial and supported submission for presentation as an interimefficacy report
Presented results reflect clinical events reported throughOctober 31, 2003
Additional non-intervention follow-up of participants is ongoing tomeet protocol requirements which may provide a more definitiveresult.
WINS: BASELINE CHARACTERISTICS
221 + 93227 + 96Time from 10 surgery toentry (SD), d
4.0%3.6% Other
8.6%10.1% Lobular
87.4%86.4% Ductal
Histologic Type
2.0 (1.6)2.0 (1.5) Mean No. + (SD)
72.9%73.1% Negative – (%)
Nodal Status
1.89 (0.9)1.93 (0.9) Mean (SD), cm
1462975Tumor Size, n
58.5 (7.61)58.6 (7.27)Age-yrs (SD)
ControlDiet
Differences between groups were not significant
WINS: BASELINE CHARACTERISTICS
7.5%6.3% AC → T
7.0%7.0% FAC/CAF
53.7%53.5% CMF
31.9%33.5% AC
763505ChemoRx Regimen, n
14.6%13.9% ChemoRx Alone
38.0%38.5% Tamoxifen + ChemoRx
47.4%47.7% Tamoxifen alone
1462975Systemic Rx, n
ControlDiet
Differences between groups were not significant
WINS: BASELINE CHARACTERISTICS
70.1%64.5% Breast Conserve
29.9%*35.5% Mastectomy
1452967Surgery, n
0.6%1.3% Borderline
29.0%28.4% Negative
67.3%67.8% Positive
1452967PgR Status, n
18.7%21.0% Negative
81.3%79.0% Positive
1462975ER Status, n
ControlDiet
* P = 0.004; other differences were not significant
Percent CAL from Fat Percent CAL from Fat
12 mo. Diet 20.3 + 7.8
p<0.0001
12 mo Control 29.2 + 8.2BaselineDiet 29.6 + 7.1
Baseline Control 29.2 + 6.7
WINS: PERCENT CALORIES FROM FAT
WINS: FAT GRAM INTAKEBY TREATMENT GROUP
56.3
57.3
51.3
33.3
53
33.4
53.7
33.9
52.2
34.8
53.9
34.8
0
30
60
Fa
t G
ram
s /
Da
y
BL 1 YR 2 YR 3 YR 4 YR 5 YR
Control
Diet
* Significantly different by T test from control and baseline, p<0.0001
* * * * *
BODY WEIGHT (POUNDS) ON STUDYBY TREATMENT GROUP
161.7 + 32.8155.6 + 32.1 *60
160.1 + 33.7157.0 + 32.9 *48
160.9 + 34.2157.1 + 32.9 *36
161.1 + 34.1157.1 + 34.1 *24
160.6 + 34.3155.6 + 33.6 *12
160.0 + 35.0 160.2 + 35.10
ControlDietMonths
All values mean + SD; a significantly different from baseline and from control, p<.0001
No. Diet = 975, 840, 781, 654, 530, 380, at O, 12, 24, 36, 48, 60, mos. respectively;No. Control = 1459, 1328, 1248, 1077, 852, 648, at 0, 12, 24, 36, 48, 60 mos, respectively.
WINS EFFICACY: RFS AND DFS EVENTS
n%n%
503.4 %282.9 %2nd Cancer1
191.3 %151.5 %Death2
25017.1 %13914.3 %DFS Events (389)
936.4 %525.3 % Distant
312.1 %111.1 %Ipsilateral Breast
372.5 %242.5 %Contra Breast
18112.4 %969.8 %RFS Events (277)
201.4 %90.9 % Local / Regional
Recurrence
Control (n=1462)Diet (n=975)
1 Other then breast cancer, 2 Without breast cancer recurrenceChlebowski, Blackburn, Elashoff, et al Proc Amer Soc Clin Oncol 24:10, 2005
Follow-up time (Years)
0 1 2 3 4 5 6 7 8
0
5
10
15
20
25
30
YEARS
ControlDiet
Absolutedifference:1%
3% 3% 3% 4% 7%
PA
TIE
NT
S (
%)
WINS: RELAPSE EVENTS
Source: Chlebowski RT et al. Presentation. ASCO 2005.
* From adjusted Cox proportional hazards model including: stratification factors, ER status,tumor size, and surgery (mastectomy/lumpectomy), p value = 0.067 by adjusted log rank test
Diet Control HR, 95% CI p-value*96/975 181/1462 0.76, 0.60-0.98 0.034
Diet 975 949 907 807 647 490 342 201 96Control 1462 1416 1352 1197 965 756 529 326 151
WINS: SUBGROUP ANALYSES
Three exploratory analyses examined dietaryeffect on relapse-free survival based on:
Body Mass Index (BMI)
Estrogen receptor status
Nodal status (positive/negative)
0 1 2 3 4 5 6 7 8
0
5
10
15
20
25
30
ControlDiet
Absolutedifference:1% 2% 1% 2% 2%
YEARS
PA
TIE
NT
S (
%)
Source: Chlebowski RT et al. Presentation. ASCO 2005.
Diet 770 753 725 641 512 385 265 156 71Control 1189 1165 1122 995 802 613 440 271 125
Diet Control HR, 95% CI68/770 122/1189 0.85, 0.63-1.14
WINS: Relapse Events in ER Positive
ControlDiet
Diet 205 196 182 166 135 105 77 45 25Control 273 205 230 203 163 133 88 55 26
0 1 2 3 4 5 6 7 8
0
5
10
15
20
25
30Diet Control HR, 95% CI28/205 59/273 0.58,0.37-0.91
Absolutedifference: 6% 8% 11% 11%6%
YEARS
PA
TIE
NT
S (
%)
Source: Chlebowski RT et al. Presentation. ASCO 2005.
P-value from adjusted Coxproportional hazard model
WINS: RELAPSE EVENTS IN ER NEGATIVE
ER and PgR RFS by Subgroups
0.85ER+, Any PgR
0.58ER-, Any PgR
0.54PgR- Any ER
0.83PgR+, Any ER
0.57ER-, PgR+
0.73ER+, PgR-
0.44ER-, PgR-
0.83ER+, PgR+
HRGroup
WINS: RFS BY BMI, NODE STATUSAND BY TREATMENT GROUP
0.83 (0.57-1.22) Positive
Node Status
0.66 (0.42-1.04) > 30
0.77 (0.56-1.07) Negative
BMI
0.77 (0.51-1.18) 25-30
0.83 (0.54-1.27) < 25
HR (95% CI)
P-value from adjusted Cox proportional hazard model
WINS: OVERALL SURVIVALBY TREATMENT GROUP
Deaths
0.4500.89, 0.65-1.2110764
P-valueHR, 95% CI
Controln=1462
Dietn=975
P-value by adjusted Cox proportional hazards modelChlebowski, Blackburn, Elashaff, et al. Proc Amer Soc Clin Oncol 24:10, 2005
No difference in overall survival
WINS: CONCLUSIONS
Dietary fat intake can be reduced in breast cancerpatients participating in a multi-center clinical trial.
A life-style intervention resulting in dietary fat reductionmay increase relapse-free survival in a population ofmostly postmenopausal breast cancer patients.
Exploratory analyses suggest a greater dietary effect inpatients with receptor negative disease
Further study of lifestyle interventions designed toimprove breast cancer outcome are warranted
Follow-up of WINS study participants continues
ACKNOWLEDGEMENTS
The women participants with breast cancer
The WINS Co-Investigators
The WINS Dieticians
The NCI for funding support
Back Up Slides
WINS: ENDPOINT ASSESSMENTPotential RFS and DFS events were identifiedat the clinical centers
Supporting medical records obtained and sentto the Statistical Coordinating Center (SCU)
The SCU removed group identifiers and sentrecords to the Clinical Director’s Officefor blinded adjudication
A final blinded adjudication was performed by WINSPathology Committee review of the records
WINS: SURGERY BY THERAPYGROUP
About 5.6% more women had mastectomy than lumpectomy in the diet comparedto control group
The surgical distribution favors the control group for local/regional recurrencesince local/regional recurrence was higher in NSABP trial in women withmastectomy compared to those with lumpectomy plus radiation 1 (14.8% versus8.1%, respectively), perhaps related to less common radiation use in themastectomy group. Since about 55 more women had mastectomy in the dietgroup, about 3 or 4 more local/regional recurrences would be anticipated in thediet group based on this imbalance (55 women x about 6.7% increased risk).
The surgical distribution slightly favors the diet group for ipsilateral recurrence.About 55 more women had lumpectomy in the control compared to diet group.Given 2.1 % ipsilateral breast recurrence in controls, about one less ipsilateralbreast recurrence would be expected in the diet group based in the imbalance.
Given these estimates, surgical imbalance does not explain the difference inbreast cancer events seen between the two treatment groups.
Fisher et al N Engl J Med 2002;347:1233