Nhìn về tương lai đối với những thử nghiệm lâm sàng hiện tại

will rapidly crunch massive amounts of data — including patient information, doctors’ notes, and results from genomics research and clinical trials — by supercomputers and cloud-based

Trang 1

Forward Looking Clinical Trials

The Great Wall International Conference on Cardiology

October 16 2014 Thach Nguyen, M.D FACC FSCAT

St Mary Medical Center

Hobart IN USA

Trang 2

1 Problems with traditional RCTs

2 New forward looking CIs

Trang 3

JUPITER

Multi-National Randomized Double Blind Placebo Controlled Trial of

Rosuvastfatin in the Prevention of Cardiovascular Events

Among Individuals With Low LDL and Elevated hsCRP

TSE 8 | 4week™, Placebo (N=8901) aria

Argentina, Belgium, Brazil, Bulgaria, Canada, Chile, Colombia, Costa Rica,

Denmark, El Salvador, Estonia, Germany, Israel, Mexico, Netherlands, Norway, Panama, Poland, Romania, Russia, South Africa, Switzerland,

United Kingdom, Uruguay, United States, Venezuela

Trang 6

= RCI is the best available technique for

eliminating confounding factors in the

assessment of a treatment effect

= With continued improvement in medical care,

most treatment effects of interest in cardiovascular disease have only modest effects (RR reductions ~15-20%)

ms Only RCI can provide sufficient precision

and confidence to reliably detect small benefits

Trang 7

= RCI’s are best suited to evaluation of

“mature” therapies

m Clinical trials are a poor way to evaluate

rapidly changing technologies, particularly

medical devices

m [rials are particularly vulnerable when

enrollment is slow or the follow-up duration is

Keyeksy

Trang 8

In a registry, all of the interventions stopped when the patients left the hospital,

Trang 9

a If the number of event too low, the patient

population may be of too low risk

Trang 11

1 Does RCT Predict the

Future?

Trang 13

2 Which Data Can

Change the Way we

Take Care of Patients ?

Trang 14

- ÈÌU J)/9 (120/09 2| (Í1[{2E©(ÌP 9/20 TÚ)

{HOw Or 60 clo Herrer’

| Deitei Selene Where revression

3] 112] D2) Lọ

Trang 15

How to Be Successful

= Business (buying stocks)

Trang 16

How to Be Successful

= Politics (who is the winner?)

The Science of

Winning Elections

Trang 18

1 Electronic health records serve as a

repository for information on millions of

people, will allow analysis of symptoms

and treatments in great numbers

Trang 19

Strategy for Success

Trang 20

Artificial intelligence algorithms

1 will rapidly crunch massive amounts of data — including patient information, doctors’

notes, and results from genomics research and

clinical trials — by supercomputers and cloud-based

programs

2 and then give out useful conclusions.

Trang 22

= [his trend toward relying on big data is

shaking up health care at every level.

Trang 23

Goals of Healthcare

= Better patient outcomes

Trang 25

Goals of Healthcare

= Decision-making based on “comparative

effectiveness,” or what has been shown to

work

Trang 26

= Medicine has always leaned on art and

intuition as well as on science But

increasingly, the art is going to be informed by serious analytics

Trang 27

1 Use mobile and web-based applications

2 Record symptoms and treatments

3 Collect data on results

4 Share the data with their providers.

Trang 28

Application la

= A patient uses a personal blood pressure monitor developed by device-maker

Ireaith

a Ihe monitor feeds BP wirelessly into an

app on his phone, and patient can transmit

it to his doctor ®

Trang 29

Application 1b

= In this case, the MD was informed that BP

was actually dangerously low The heart

rate was checked and was too low so

patient ended up with a pacemaker

Trang 30

= | Replace the general guidelines often

used in deciding how to treat unstable

angina, AMI, or Heart failure

Trang 32

could make up-to-date predictions about

the individual patient a doctor is seeing.

Trang 35

= 500 patients with HF in NYHA FC Ill or IV

= Follow-up from 2002 to 2004

= Ihe package survival available in the R

software was used to obtain the results for

model (1) Results for models (2) and (3)

were obtained in this same software using

the packages timereg and Coxvc, as well

as for models (4) and (5) using the package timereg.

Trang 36

Table 2 Logrank test performed for each

CO Yall tally

Logrank test

Age (=60 and >60 yrs) 13.00 <0.001 Gender (male and female) 2.64 0.104 Race (white and others) 0.56 0.453 Diabetes mellitus (yes and no) 1.04 0.307

Hypertension (yes and no) 1.54 0.215

Current smoking (yes and no) 1.96 0.165 BMI (<25 and >25 kg/m2) 6.16 0.013

LV ejection fraction (<0.35 and =0.35) 10.70 0.001

LV mass (=243 and >243 g) 0.11 0.742

Serum sodium (=137 and >137 mEq/L) 27.9 <0.001 Hemoglobin (Hb) (=13 and >13 g/dL) 15.6 <0.001 Creatinine (<1.2 and >1.2 mg/dL) 23.4 <0.001 Etiology (Chagas and others) 13.13 <0.001

doi:10.137 1/journal.pone.0037392.t002

Giolo SR, Krieger JE, Mansur AJ, Pereira AC (2012) Survival Analysis of Patients with Heart Failure: Implications of Time-Varying Regression Effects

in Modeling Mortality PLoS ONE 7(6): e37392 doi:10.1371/journal.pone.0037392

Trang 37

Table 4 Tests associated with the additive

*Time-invariant effects suggested for age and serum sodium (p> 0.05)

All covariates were centered in their respective mean values

do&10.1371/journal pone.0037392.t004

Trang 38

Figure 3 Cumulative coefficients obtained

from the additive hazards model

Trang 39

Criminals and Fugitives Identified by Robotcop Glasses

Trang 40

Hidden Adversaries Located by

Trang 41

A whole hospitalization course, cost,

date of discharge could be generated

when the pafient checks In

Trang 42

Asymptomatic patients could be scanned

for future relapses and so the problems

Trang 44

CONCLUSIONS

Trang 45

The new draft guidance, on the basis of the

EXCELLENT and PRODIGY studies, recommend a reduced duration of DAT after

DES: six months, instead of 6 to 12

The draft guidelines also specify, however, that

“more than six months" is preferable in patients at high ischemic risk and low bleeding

We

Trang 46

HR (95% CI)=

0.95 (0.80—1.1) P=0.51

Trang 47

Study in NVAF patients undergoing PC!

endpoint (PROBE) study to Eva lua te DUAL antithrombotic thera py with dabeatr n etexibte (11Ong h.id.a nd 150mg b.i.d.] re c lo ndce rel or

bf tte es) es each eee al (INR 20—-3.0 pluschpdogrelorticagrelor with aspirin in patients with non valvularatral

†ibril tien (Nv ÀF] that ha+e underponea

D1 50 plus a P2Y12 inhibitor is:

Nor-infe nor with respect tothe

combined thrombotic event rate [TE:

death+ Ml+stmle/SE]

0|,

N¬arr-irfe ñnar? 0ith spectto clinically mlevant blaeding relitive to

a tripk combiration of arfarin plus

DiiO plusa P2¥12 inhibitor 6:

Non- inferior with respect to the

co mì b ned thro bat &¿e nt ra tế (TE:

- C0 10214204 13,

AND

relsva mt hl£edinp reb tiee to a trí rk=

combination of warfarin plusa P2V12

Pe'cutaneous corona ry inte rvention (PCI) wíth inhibitor (clo pidogrelor ticagrelor) plus a P2¥12 inhibitor (clopidogrel or

stenting (RE-OUAL FOI) <

^ ị oan ticagrelor) plus ASA

Paroxysmal, persistent or

permanent AF

(PCI with stenting [BMS

or DES] elective or ACS)

Dabigatran 150mg BID + P2¥12 inhibitor¥**

dbo Soc

Time to first combired thrombotic event ordeath

falldeath, MI, Stroke/SE)

Dabigatran 110mg BID + P2¥12 inhibitor¥**

Screening

0-72 hours post-PC! Plus

Time to first clinica lh reevant bleeding rate

18/24/30M (ISTH Majp r)

bm 8M 12M

or EOT

n = 2840 patients perarm (Total = 8520 patients)

|

"(G8 willbe decontimed inthe warferinerm 8hS: Decontinustionof (GA st month 4 ; DES: d&continustionof (GA at month?

orrelor Ticsgzretor, The P2412 inhibforcen t decontiwedoefter month 12 offolkay upstthe decretonot the p

Trang 48

Figure 3 Cumulative coefficients obtained

from the additive hazards model

Trang 49