transplant benefit-based liver allocation

SRTR

Transplant Benefit-BasedLiver Allocation

Robert M. Merion, MD, FACS

OPTN/UNOS Liver ForumAtlanta, GA

April 12, 2010

SRTR

Outline

• Introduction: organ allocation schemes• Transplant Benefit Score• Comparison to MELD Score• Simulation results: LSAM

– Comparison of deaths, life-years saved, and distance• Benefit score vs. MELD• Benefit score vs. geographic modifications of MELD

• Conclusions

SRTR

Background

• Currently, chronic liver failure patients waitlisted for deceased-donor liver transplantation are prioritized with respect to medical urgency in decreasing order of Model for End-stage Liver Disease (MELD) score

• Shortfall in donor livers increases pressure to make the best possible use of available organs

• It has been suggested that post-transplant survival should play a role in liver allocation

• Transplant benefit has been a central component of the allocation system since May, 2005.

SRTR

Organ Allocation Schemes

SRTR

Organ Allocation Policy

• Possible bases for organ allocation:

1. URGENCY: future wait-list lifetime

2. UTILITY: post-transplant lifetime

3. BENEFIT: combines (1) and (2)

SRTR

Organ Allocation: Example Expected Lifetimes

PatientID #

PredictedWL

Lifetime

Predictedpost-LTLifetime

1 7 10

2 2 3

3 5 9

SRTR


• Q: An organ is procured. To which patient should it be allocated? Patient

ID #

PredictedWL

Lifetime


1 7 10

2 2 3

3 5 9

SRTR


• Q: An organ is procured. To which patient should it be allocated?

• A: depends on the allocation rules …..

PatientID #

PredictedWL

Lifetime


1 7 10

2 2 3

3 5 9

SRTR


• Minimum WL lifetime:– Allocate to: #2

PatientID #

PredictedWL

Lifetime


1 7 10

2 2 3

3 5 9

SRTR



• Maximum post-LT lifetime:– Allocate to: #1

PatientID #

PredictedWL

Lifetime


1 7 10

2 2 3

3 5 9

SRTR



• Maximum post-LT lifetime:– Allocate to: #1

• Benefit (LT - WL):– Allocate to: #3

PatientID #

PredictedWL

Lifetime


1 7 10

2 2 3

3 5 9

SRTR

Organ Allocation: Example

ID#

Future Lifetime (patient)

WL LT Benefit =LT-WL

1 7 10 3

2 2 3 1

3 5 9 4

SRTR

Transplant Survival

Benefit Score

SRTR

Liver Transplant Survival Benefit

• Concept developed in collaboration with OPTN/UNOS Liver Committee since 2006

• Patient-specific and donor-specific• Separate models for waitlist and post-transplant lifetimes• Uses available factors other than MELD components• Difference in area under 5-year survival curves

– Predicted post-transplant lifetime minus predicted future waitlist lifetime

– Reflects life years gained through liver transplantation

SRTR

Waitlist Survival Model: Covariates

• Not just MELD• Creatinine, bilirubin, INR, albumin, sodium, dialysis,

age, BMI, diagnosis, HCC, diabetes, hospitalization status, prior malignancy, growth failure, previous time on waitlist, rate of change: creatinine, bilirubin, albumin

SRTR

Post-Transplant Survival Model: Covariates• Not just MELD• Recipient factors:

– Creatinine, albumin, age, diagnosis, diabetes, dialysis, hospitalization status, previous liver transplant, life support, portal vein thrombosis, previous abdominal surgery, hepatitis C, growth failure

• Donor factors: – Age, race, cause of death, donation after cardiac death

SRTR

Waitlist Survival

SRTR

Waitlist 5-Year Expected Lifetime

SRTR

Post-Transplant Survival

SRTR

Post-Transplant 5-Year Expected Lifetime

SRTR

Mean 5-Year Future Lifetime by MELD

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

MELD

Life

expec

tancy

(ye

ars)

-

Transplant

Benefit

post-LT

WL

SRTR

Transplant Benefit by MELD

-2.0

-1.0

0.0

1.0

2.0

3.0

4.0

6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

MELD

5-ye

ar B

enef

it (ye

ars)

-

SRTR

Transplant Benefit vs. MELD• Q: If average transplant benefit increases with MELD, does

allocation by benefit amount to allocation by MELD?

SRTR


allocation by benefit amount to allocation by MELD?• A: NO!

SRTR



– e.g., two patients could have similar (or equal) MELD scores, but very different benefit scores

SRTR



– e.g., two patients could have similar (or equal) MELD scores, but very different benefit scores

– MELD is not the only factor predicting waitlist and post-transplant survival

SRTR

5-Year Transplant Benefit by MELDBox-Whisker Plots

Rank correlation between MELD and transplant benefit score = 0.67

SRTR

Calculation of MELD and Benefit Scores

SRTR

MELD: CalculationVariable Coefficient

Intercept 0.6431

Not on dialysis:log [min{max(Creatinine,1),4}]

0.957

log (max{Bilirubin,1}) 0.378

log (max{INR,1}) 1.120

On dialysis 1.326

• Notes: – Multiply the score by 10 and round to the nearest whole number.– MELD score is bounded by 6 and 40 – HCC T2 candidates receive an exception score of 22.

SRTR

PELD: Calculation

• Notes: – Multiply the score by 10 and round to the nearest whole number. – Laboratory values less than 1.0 are set to 1.0 for the purposes of the

PELD score calculation.

Variable Coefficient

log (Bilirubin) 0.480

log (INR) 1.857

log (Albumin) -0.687

Patient is less than 1 year old 0.436

Growth failure 0.667

SRTR

Benefit Score: Calculation

• Calculation of benefit score by linear approximation is analogous to MELD score– more terms

• Comparison to statistically rigorous calculation– almost perfect correlation (rank correlation 0.99)– used in simulations to reduce computing time

• Linear approximation would be utilized by OPTN/UNOS if incorporated into a future allocation system, just as MELD is currently

SRTR

Microsimulation to Compare Allocation

Systems

SRTR

Requests by OPTN/UNOS Liver Committee

• LSAM Simulations Study Population – Data from waitlist candidates and donors during 2006 were

used for the simulations.• Compared multiple broader distribution, MELD-based

and transplant benefit-based allocation systems to the current allocation system

• For each run, we recorded: number of transplants, mean transplant benefit, life-years saved, total deaths, median distance traveled, percent shared, and percent distance traveled >100 nautical miles.

• Results averaged over 10 iterations

SRTR

InitialWaitlist

NewCandidates

DonatedOrgans

Wait List Post-graft

Survival

RemovalFrom

Waitlist

OrganAllocation/ Placement

WaitlistMortality

Relisting

Post-Transplant Mortality

Object Pascal SAS

LSAM Event-Sequenced Modeling

Waitlisted Patients’ histories

SRTR

LSAM: Number of Deaths

Deaths MELD/PELD System

Regional Sharing

Transplant Benefit

Waitlist 1,6601,602(-58)

1,519(-83)

Post-transplant 609

607(-2)

601(-6)

Post-removal 407397(-10)

384(-13)

Total 2,6752,606(-69)

2,504(-102)

8%

6%

(n) Indicates difference vs. prior column; n% indicates difference vs. MELD/PELD system

SRTR

LSAM: Life Years Gained by Transplant

MELD/PELD System

Regional Sharing

Transplant Benefit

Mean Extra Years From Transplant

1.561.63

(+0.07)2.01

(+0.38)

∑ Life Years Gained by Transplant

9,87510,225(+350)

12,448(+2,223) 23%

(n) Indicates difference vs. prior column; n% indicates difference vs. MELD/PELD system

SRTR

Equity: Age of Transplanted Patients

SRTR

Equity: Gender of Transplanted Patients

SRTR

Equity: Race/Ethnicity of Transplanted Patients

SRTR

Comparison of Allocation Rules for Adult Deceased Donor Livers

Yellow = UnchangedBlue = Current SystemGreen = Share Positive Benefit

Current SystemLocal – Status 1ARegional – Status 1ALocal – Status 1BRegional – Status 1BLocal – MELD/PELD 15Regional – MELD/PELD 15Local – MELD/PELD < 15Regional – MELD/PELD < 15National – Status 1ANational – Status 1BNational – MELD/PELD

Share Positive BenefitLocal – Status 1ARegional – Status 1ALocal – Status 1BRegional – Status 1BLocal – Transplant benefit score > 0Regional – Transplant benefit score > 0Local – Transplant benefit score < 0Regional – Transplant benefit score < 0National – Status 1ANational – Status 1BNational – Transplant benefit score

SRTR

Decrease in Total Deaths (vs. Current)

51

0

51 46

163

61

112

147

0

25

50

75

100

125

150

175

200

Share PositiveBenefit

Current Share 25L Share 25R RegionalSharingBenefit

RegionalSharing

MELD/PELD

Concentric 25 Five Zones

Num

ber

of T

otal

Dea

ths

-

SRTR

Increase in Life Years Saved via Transplant (vs. Current)

0

500

1,000

1,500

2,000

2,500

3,000

SharePositiveBenefit


RegionalSharing

MELD/PELD

Concentric25

Five Zones

Additio

nal

Life

Yea

rs S

aved

-

SRTR

Median Distance Between Donor Hospital and Transplant Center

0

50

100

150

200

250

SharePositiveBenefit


RegionalSharing

MELD/PELD

Concentric 25 Five Zones

Dis

tance

(nau

tica

l m

iles

) -

SRTR

Median Distance vs. Decrease in Total Deaths

0

50

100

150

200

0 50 100 150 200 250

Median Distance (nautical miles)

Dec

reas

e in

tota

l dea

ths

-

current

Shr Pos Benefit

Five Zones

Concentric 25

Reg Shr Benefit

Reg Shr MELDShr 25L

Shr 25R

SRTR

Median Distance vs. Decrease in Total Deaths

Current

Shr Pos Benefit

Five Zones

Concentric 35

Reg Shr Benefit

Reg Shr MELD

Shr35R, National 15

250_25250_35 250_22

250_29National

15/3515/32

15/29 15/2515/22

Concentric

Concentric

250_1535

250_1532 250_1529250_1525

250_1522MELDNaRegShr

MELDNa

SRTR

Percent Shared vs. Decrease in Total Deaths

Current

Shr Pos Benefit

Five Zones

Concentric 35

Reg Shr Benefit

Reg Shr MELD

Shr35R, National 15

15/3517/35

20/35

Share

250_25250_35 250_22250_29

Concentric

MELDNaRegShr

MELDNa

SRTR

Benefit Scores, Life-Years Saved, and Distance Traveled

Allocation SystemMean

BenefitScore

Life-YearsSaved in 1

Year

Average Distance

Regional Sharing Benefit 2.01 12,448 124

Share Positive Benefit 1.87 11,794 59

Regional Sharing MELD/PELD 1.63 10,217 127

Current 1.57 9,929 65

SRTR

Benefit Score: Reducing Complexity

• Can a modified benefit score be calculated with fewer factors?

• This work is ongoing, and decisions about the tradeoffs can be made on clinical and statistical grounds by examining the effect on the rankings of candidates under various scenarios with fewer factors

SRTR

Summary

• Incorporation of a measure of post-transplant survival into the allocation system is being actively explored

• No specific policies or proposals have been put forward

• The role of transplant benefit score as a criterion for liver allocation will require further discussion

• Input from the transplant community is extremely important

transplant benefit-based liver allocation

Documents