Standard Traditional Meta-Analysis Different times since start of treatment Mantel-Haenszel approach.

Standard Traditional Meta-Analysis

•Different times since start of treatment

•Mantel-Haenszel approach

Analysis of a Single Time Interval (1)

Treatment Deaths Survivors Total

A

B

Total

6

5

11

20

24

44

26

29

55

Treatment A: Observed deaths = 0 = 6

Expected deaths = E = (11 × 26)/55

= 5.200

Variance = V = (11 × 44 × 26 × 29)/552 × 54) = 2.234

Odds ratio = OR = (6/20)/5/24)

= 1.440

Log (OR) = LOR = 0.365

Analysis of a Single Time Interval (2)

Log (OR) = 0.365

LORapprox = (0 – E)/V = 0.358

Variance (LOR) = 1/V = 0.448

SD = 0.669

95% CI (LOR) = (0.358 – 1.96 × 0.669)

to

(0.358 + 1.96 × 0.669)

= (- 0.953, 1.669)

95% CI (OR) = (0.38, 5.41)

Analysis of Multiple Time Intervals

Time Interval 0 E V

1

2

3

:

01

02

03

:

E1

E2

E3

:

V1

V2

V3

:

Meta-analysis 0 E V

log (OR) = (0 - E)/V

etc.


•Different hospitals within a trial


•Checking the assumptions: is the hazard ratio between treatments constant between different circumstances (time since treatment began, prognostic factors, hospitals, etc)?

•Power to test these assumptions is generally weak - need biological insight: e.g. ER+ve vs ER-ve

(1) Make bold, simplifying assumptions; look at as wide arange of data as possible; be critical but notpernickety. Do not be disturbed if the data does notexactly fit the model.

(2) It is important not to place too much reliance on anysingle piece of evidence, since some data is bound tobe misleading if not plain wrong.

(3) Make no assumptions that can not be doubted fromtime to time.

(4) Almost all aspects of life are engineered at themolecular level, and without understanding moleculeswe can only have a very sketchy understanding of lifeitself.

See Francis Crick "What Mad Pursuit".

‘New’ Forms of Meta-Analysis

•Different trials

•Different trials with ‘slightly’ different regimens

•Different trials with ‘slightly’ different definitions of prognostic factors that influence drug response

‘New’ Forms of Meta-Analysis

•What contributes to differences between trials?

•Regimens used?

•General health of patients?

•Support services?

Increase in Breast Cancer Risk with Use of ERT and EPRT

Increase in risk

Study HRT Type per 5 years of use

Magnusson et al.a ERT 16%

EPRT 40%

Schairer et al.a ERT 1%

EPRT 40%

Ross et al.c ERT 6%

EPRT 24%

aInt J Cancer 1999; 81:339-344.bJAMA 2000; 283:485-535.cJ Natl Cancer Inst 2000; 92:328-332.

B056-2

Hawaii/Los Angeles MEC Study: HRT use

HRT W AA NH JA L-US L-NUS

Never 36% 56% 54% 40% 50% 63% Past HRT

20% 23% 18% 14% 20% 20%

Current ERT

14% 11% 11% 15% 12% 7%

Current EPRT

30% 10% 19% 30% 18% 10%

Age (years)

30 40 50 60 70

Inc

ra

te /

10

0,0

00

1

10

100

500

Breast Cancer

B1

Day of Cycle

Estradiol (pg/ml)

Progesterone (ng/ml)

Day of Cycle

0

0.5

1

1.5

2

2.5

1 3 5 7 9 11 13 15 17 19 21 23 25 271 5 7 9 13 17 19 23

Relative Labelling Index

3 5 7 9 11 13 15 17 19 21 23 25 271

400

300

200

100

0 0

5

10

15

20

Breast Cell Proliferation

B051

Breast cell mitotic rate on continuous-combined estrogen-progestin replacement therapy is more than double that of estrogen replacement therapy.

Hofseth et al. (JCEM 1999; 84:4559-4565).

HRT and Mammographic Densities: BI-RADS Changes in the PEPI Trial*

BI-RADS Categorization: (1) Entirely fatty (2) Fatty with scattered fibroglandular tissue (3) Heterogeneously dense (4) Extremely dense

Women with an increase in Estrogen† Progestin‡ BI-RADS Category after 12 months ________________________________________________________________________

Placebo Placebo 0% CEE (0.625 mg/d) Placebo 3.5% CEE (0.625 mg/d) MPA (10 mg/d; 12 days/mo) 23.5% CEE (0.625 mg/d) MPA (2.5 mg/d) 19.4% CEE (0.625 mg/d) MP (200 mg/d; 12 days/mo) 16.4% *Greendale et al. (Ann Int Med 1999; 130:262-269). †CEE = conjugated equine estrogens. ‡MPA = medroxyprogesterone acetate. MP = micronized progesterone.

B123

PREMPRO

0.625 mg CE/2.5 mg MPA

ORTHO-PREFEST

1 mg E2/0.09 mg norgestimate for 3 days every 6 days

FEMHRT

5 µg EE2/1 mg norethindrone acetate

H026

Approaches to Reducing Breast Progestin Exposure with HRT

1. Use of progestin for 12-14 days every 3 months.

Ettinger et al. (Obstet Gynecol 1994; 83: 693-700).

Williams et al. (Obstet Gynecol 1994; 84: 787-793).

E060


2. Use an intra-uterine device - delivers local progestin

with little systemic effect. Shoupe et al.

(N Engl J Med 1991; 325: 1811-1813).

E060


3. Use of intra-vaginal route of delivery - 50-fold gain in endometrial concentration for equal blood level of progesterone.

Miles et al. (Fertil Steril 1994; 62:485-490).

Fanchin et al. (Obstet Gynecol 1997; 90: 396-401).

E060

Meta-Analysis or Large Trial Problems

•The study does not address the right question.

E.g. the WHI uses EPRT as the HRT arm.

TR034

Age (years)

20 30 40 50 60 70 80

Inc

rate

/ 10

0,00

0

100

200

300

400

500

600

1973-78

1991-96

First author % fat calories

(year) Intervention

PremenopausalWoods (1989) 25

Hagerty (1988) 25

Rose (1987) 21

Boyd (1997) 21

Williams (1989) 20

Goldin (1994) 20

Woods (1996) 20

Ingram (1987) 18

Schaefer (1995) 18

Bogga (1995) 12

PostmenopausalCrighton (1992) 24

Prentice (1990) 20

Ingram (1987) 18

Heber (1991) 10

All studies

All above excluding Bagga (1995) and Heber (1991)

0 0.5 1 1.5 2

Estradiol level (relative to baseline) Hor018

Hormonal Breast Cancer Chemoprevention in Premenopausal

Women

How do we prove that breast cancer prevention

will occur?

(a) Randomized trials?

(b) Common sense?

Mammographic changes?

Breast cell proliferation?

(c) Show that the proposed chemopreventive

regimen is an ‘acceptable’

contraceptive? This route means that

the regimen has to be inexpensive.


Women

Can we maintain the protection against ovarian

cancer and endometrial cancer afforded by

the Pill?


Women

Menopausal symptoms

QOL

Bone

CVD

StrokeDVT

Mode of delivery: oral; patch; intra-nasal

Hormonal Breast Cancer Chemoprevention in Premenopausal Women

How is this development going to be paid for?

Our approach: Deslorelin + Add-back E2TP4 :

Treat uterine fibroids - compare to current treatment

Treat endometriosis - compare to current treatment

Treat PMS?

Hormonal contraception?

Current studies:

Can bone be preserved

while bleeding is controlled in fibroids

while endometriosis is controlled

What does the FDA Demand?

• GNRH agonist plus add-back– Estradiol + Testosterone + Progesterone– Only permitted to do Estradiol

Risk of Breast Cancer and Postmenopausal

Serum Estradiol Concentrationa

Serum E2 Concentration (pmol/L)

Odds Ratio (with 95% CI)

<30.7 1.0 30.7-41.0 2.5 (1.0-6.4)

>41.0 5.0 (2.0-12.5)

(p < 0.001)

aThomas et al. (Br J Cancer 1997; 76: 401-405).

O028

Intra-Ovarian Estradiol and Ovarian Cancer Risk

The observed 25% lower serum E2 in Japanese women is predicted on

the basis of in vitro results to produce a 3.2-fold reduction in their ovarian cancer rate cf U.S. women; decreasing to 4.2 when adjustment is made for earlier age at menarche and decreased parity.

Great need is find out what is driving cell proliferation in vivo. Only in this way will we be sure that new formulations of hormonal contraceptives will not lose the chemopreventive benefits of OCs. For example, does Depot Provera prevent ovarian cancer?

O041

Ovarian Cancer

• Mitogens for ovarian ‘epithelium’: FSH and E2 (at an intra-ovarian level) - E2 action blocked by P4

• Needs: what are cell proliferation changes in the menstrual cycle? Study cystadenomas?


•Adjusting for prognostic factors


•Adjusting for prognostic factors

•Cox Proportional hazard approach


•What is the final survival curve comparison?

10 30 50 70 90

0

20

40

60

80

GN028

Standard Traditional Meta-Analysis Different times since start of treatment Mantel-Haenszel approach.

Documents

e060 slide

trial slide

b123 slide

breast cancer b1 slide

drug response slide

use of progestin

breast progestin exposure

er ve vs erve slide