LFDAAEDM

ptulwttmws

pMfse

IutsIeoidesmvtetgr

A

PS

ACC

©P

GEN

ERA

LT

HO

RA

CIC

ung Transplantation for Idiopathic Pulmonaryibrosisavid P. Mason, MD, Mariano E. Brizzio, MD, Joan M. Alster, MS,nn M. McNeill, RN, Sudish C. Murthy, MD, PhD, Marie M. Budev, DO,tul C. Mehta, MD, Omar A. Minai, MD, Gösta B. Pettersson, MD, PhD, andugene H. Blackstone, MDepartments of Thoracic and Cardiovascular Surgery, Quantitative Health Sciences, and Pulmonary, Allergy, and Critical Care

edicine, Cleveland Clinic, Cleveland, Ohio

I5pdtSwys7

ItS

Background. Outcomes of lung transplantation for idio-athic pulmonary fibrosis (IPF) are thought to be worse

han those for other indications, although the reasons arenknown. In addition, the choice of single versus double

ung transplantation is unclear. To guide decision-making,e (1) compared survival of patients receiving transplanta-

ion for IPF with survival of patients receiving transplan-ation for non-IPF diagnoses, (2) identified risk factors for

ortality after transplantation for IPF, and (3) ascertainedhether double lung transplantation for IPF confers a

urvival advantage.Methods. From February 1990 to November 2005, 469

atients underwent lung transplantation, 82 for IPF.ultiphase hazard modeling was used to identify risk

actors, and propensity matching was used to compareurvival of IPF and non-IPF patients and to assess the

ffect of single versus double lung transplantation.

sTavttba

P

PFultcSd

fabrr

ardiovascular Surgery, Cleveland Clinic, 9500 Euclid Ave, Desk F24,leveland, OH 44195; e-mail: masond2@ccf.org.

2007 by The Society of Thoracic Surgeonsublished by Elsevier Inc

Results. Survival estimates after transplantation forPF were 95%, 73%, 56%, and 44% at 30 days and 1, 3, and

years, somewhat worse than for matched non-IPFatients (p � 0.03). Risk factors for mortality were earlierate of transplantation (p � 0.07), single lung transplan-

ation (p � 0.03), and higher wedge pressure (p � 0.003).urvival for double versus single lung transplantationas 81% versus 67% at 1 year and 55% versus 34% at 5

ears; however, among matched non-IPF patients, corre-ponding survivals were 88% versus 71% at 1 year and2% versus 48% at 5 years (p � 0.3).

Conclusions. Survival after lung transplantation forPF is worse than after other indications for transplanta-ion when multiple clinical variables are accounted for.urvival may be improved by double lung transplant.

(Ann Thorac Surg 2007;84:1121–8)

© 2007 by The Society of Thoracic Surgeons

diopathic pulmonary fibrosis (IPF) is the most commonadult form of diffuse parenchymal lung diseases of

nknown origin and is the second most common indica-ion for transplantation in the United States after emphy-ema. Transplant outcomes are thought to be worse forPF than for other indications, although there is noxplanation for this difference [1–4]. However, the largestbservational studies demonstrating a high early mortal-

ty after transplantation for IPF are not well controlled foronor, recipient, and transplant variables [5]. We hypoth-size that survival after transplantation for IPF might beimilar to that for other indications if early death could beinimized and patients matched on important clinical

ariables. In addition, choice of single versus double lungransplantation for IPF remains debated [6]. We hypoth-size that patients undergoing double lung transplanta-ion for IPF might have better survival than those under-oing single lung transplantation, as seen in patientseceiving lung transplants for emphysema, if similar

ccepted for publication April 23, 2007.

resented at the Poster Session of the Forty-third Annual Meeting of Theociety of Thoracic Surgeons, San Diego, CA, Jan 29–31, 2007.

ddress correspondence to Dr Mason, Department of Thoracic and

hortcomings of early death could be minimized [7, 8].herefore, to guide future decision-making, we evalu-ted our single-center experience and (1) compared sur-ival of IPF versus non-IPF patients using matched pa-ient populations, (2) identified risk factors for death afterransplantation for IPF, and (3) ascertained whether dou-le lung transplantation for IPF confers a survivaldvantage.

atients and Methods

atientsrom February 1990 to November 2005, 469 patientsnderwent primary lung transplantation for end-stage

ung disease at Cleveland Clinic, exclusive of heart–lungransplantation. Eighty-two patients (17%) fulfilled theriteria for IPF as outlined by the American Thoracicociety [9]. Thirty-two (39%) of the 82 IPF patients un-erwent double lung transplantation.Recipient, donor, and surgical data were extracted

rom the Unified Transplant Database, which has beenpproved for use in research by the institutional reviewoard, with patient consent waived. The institutionaleview board approved supplemental review of medical

ecords, also with patient consent waived.

0003-4975/07/$32.00doi:10.1016/j.athoracsur.2007.04.096

ETmff(t5afwo

DS

N

natw

mahsfivvoaa

S

amiesovllawtoetctsmcippaps

O

dttArp0

R

IPRwaanhactyw

irtUm

RIRewrp

SIDmwypptvswtfsbs

1122 MASON ET AL Ann Thorac SurgLUNG TRANSPLANTATION FOR IPF 2007;84:1121–8

GEN

ERA

LT

HO

RA

CIC

nd Pointhe primary end point was all-cause mortality. Althoughost patients are followed clinically at Cleveland Clinic,

ormal anniversary follow-up is obtained yearly. Meanollow-up of surviving IPF patients was 3.9 � 2.8 yearssingle lung transplant, 3.3 � 1.4 years; double lungransplant, 4.5 � 3.7 years); 18% were followed more than

years, and 227 patient-years of data were available fornalysis. Among all patients (IPF and non-IPF), meanollow-up for surviving patients was 4.2 � 3.0 years; 32%ere followed more than 5 years, and 1,565 patient-yearsf data were available for analysis.

ata AnalysisURVIVAL AFTER LUNG TRANSPLANTATION FOR IDIOPATHIC PULMO-

ARY FIBROSIS. Risk-unadjusted survival was estimatedonparametrically by the method of Kaplan and Meier,nd parametrically by a multiphase hazard decomposi-ion method [10]. (For additional details, see http://ww.clevelandclinic.org/heartcenter/hazard.)Nonproportional, multiphase, multivariable hazardethodology [10] was used to identify recipient, donor,

nd procedure variables (Appendix) associated with eachazard phase simultaneously. The process began withcreening of values for variables to ensure that at leastve IPF deaths were associated with each dichotomousariable. Bootstrap aggregation (bagging) was used forariable selection [11, 12] with a probability for inclusionf 0.05; variables appearing in at least 50% of bootstrapnalyses were considered reliably statistically significantt a probability less than 0.05 (median rule).

URVIVAL COMPARISON. Although survival may be worsefter lung transplantation for IPF versus non-IPF, IPFay be simply a marker for a high-risk group of patients

ndependent of the cause of their pulmonary disease. Forxample, IPF patients were more likely to be male,omewhat older, and larger than those with other etiol-gies (Table 1). As expected, their forced expiratoryolume in 1 second was greater, but forced vital capacityess. Therefore, a propensity score for IPF was formu-ated, based on logistic regression of 24 donor, recipient,nd transplant variables (Appendix) [13, 14]. The goalas to identify two subgroups of IPF and non-IPF pa-

ients matched as closely as possible so that comparisonf survival of those patients would reflect a true differ-nce rather than disparities in recipient and donor fac-ors. Thus, variables were chosen that were generalharacteristics of patients undergoing lung transplanta-ion, but not those that were surrogates for IPF (such aspirometry values). The C-statistic for the propensityodel was 0.84. From this model, a propensity score was

alculated for each patient and used for matching, yield-ng 70 IPF and non-IPF matched patient pairs. Twelve IPFatients were unable to be matched to similar non-IPFatients. Using the data set of matched patients, risk-djusted comparison of survival for IPF versus non-IPFatients was performed. In addition, characteristics and

urvival of unmatched patients were explored (Table 2). l

NE LUNG OR TWO? In addition to multivariable analysis ofeath that included single versus double lung transplan-

ation, another propensity analysis was performed withinhe IPF group for single versus double lung transplant.ll variables described in the Appendix (including spi-

ometry values) were considered in this analysis. Aropensity score based on seven variables (C-statistic,.94) yielded 10 well-matched pairs.

esults

diopathic Pulmonary Fibrosis Versus Non–Idiopathiculmonary Fibrosis Survival Comparisonisk-unadjusted survival after transplantation for IPFas 95%, 73%, 56%, 44%, and 36% at 30 days and 1, 3, 5,

nd 7 years, which was similar to 95%, 81%, 61%, 48%,nd 38% at these same intervals after transplantation foron-IPF indications (p[early hazard phase] � 0.6, p[lateazard phase] � 0.5; Fig 1A). Instantaneous risk of deathfter lung transplantation for both IPF and non-IPFonsisted of two hazard phases that differed early, but fello equivalent levels after about 6 months, with risk at 2ears of 13% per year (Fig 1B). Recipient causes of deathere similar in IPF and non-IPF patients.Among propensity-matched patients, those undergo-

ng transplantation for IPF experienced somewhat worseisk-adjusted survival than those undergoing transplan-ation for other indications (p � 0.03, log-rank test; Fig 2).nmatched patients had survivals similar to those ofatched patients in both groups.

isk Factors for Death After Transplantation fordiopathic Pulmonary Fibrosisisk factors amplifying the early phase of risk includedarlier date of transplantation for IPF. Late risk of deathas increased by higher wedge pressure measured at

ight heart catheterization as well as single lung trans-lantation, irrespective of side (Table 3).

ingle Versus Double Lung Transplantation fordiopathic Pulmonary Fibrosisouble lung transplantation for IPF was initially uncom-on, then increased in frequency, and now has some-hat decreased (Fig 3). It tended to be performed on

ounger patients (p � 0.0001; Fig 4). Thus, only 10atients could be matched to single-lung-transplant IPFatients to explore whether single lung transplant was a

rue risk factor or a marker for high-risk patients. Sur-ival after either double or single lung transplant in thesemall subgroups of matched patients was considerablyorse than that of unmatched patients (Fig 5), indicating

hat the risk factor “single lung transplant” was a markeror a high-risk subgroup of older patients with IPF. Thisimilar survival in matched patients (p � 0.3) was verifiedy forcing propensity score into the multivariable analy-is of Table 2 and observing that the p value for single

ung transplant rose to 0.3.

C

Lsawstfnwtip

tooovrfh

PS

N

T

C

R

D

S

a

F acityP

1123Ann Thorac Surg MASON ET AL2007;84:1121–8 LUNG TRANSPLANTATION FOR IPF

GEN

ERA

LT

HO

RA

CIC

omment

ung transplantation for IPF has been shown to confer aurvival benefit over best medical therapy, but survivalppears to be worse than that for non-IPF indicationsithout clear explanation [2, 15]. In addition, whether or not

urvival is worse with single versus double lung transplan-ation remains unclear. Although multi-institutional studiesocusing on these questions have benefited from largeumbers for statistical analysis, they have included centersith varied experience with transplantation for IPF, and

his may account for the high early mortality [5]. Compar-son accounting for multiple donor, recipient, and trans-

able 1. Recipient, Donor, and Procedure Characteristics of P

haracteristic

IPF Pat(n �

na No. (%)

ecipientDemography

Age (y) 82Height (cm) 82 1Women 82Caucasian 82

ComorbiditiesHistory of smoking 63

ImmunologyPRA �10% 82

Cardiac and pulmonary measuresFEV1 (% of predicted) 62FVC (% of predicted) 62Wedge pressure (mm Hg) 62PA systolic pressure (mm Hg) 69PA mean pressure (mm Hg) 67

Cause of posttransplantation deathRespiratory infection or failure 40Sepsis or multisystem organ failure 40Malignancy 40Other 40

onorDemography

Age (y) 80Height (cm) 82 1Women 82Caucasian 82

Lung sizeTotal lung capacity (L) 80 5

Cause of deathHead trauma 82

urgical procedureDouble lung transplantation 82Years from 1/1/1990 to transplantation 82Cold ischemic time (min)b 69 2

Data available. b For double lung transplantations, maximum of left

EV1 � forced 1-second expiratory volume; FVC � forced vital capRA � panel-reactive antibody; SD � standard deviation.

lant variables is difficult in multi-institutional studies. For h

his reason, we evaluated our single-institution experiencef transplantation for IPF with careful propensity matchingf patients to determine whether IPF patients have worseutcomes than non-IPF patients after important clinicalariables are carefully controlled. In addition, we identifiedisk factors for death after transplantation for IPF andocused on whether single or double lung transplantationeld a survival advantage.

rincipal FindingsURVIVAL AFTER LUNG TRANSPLANTATION FOR IDIOPATHIC PULMO-

ARY FIBROSIS. Patients undergoing transplantation for IPF

ts Undergoing Lung Transplantation

Non-IPF Patients(n � 387)

p Valueean � SD na No. (%) or Mean � SD

11 387 47 � 13 0.00069.9 387 167 � 11 0.004) 387 204 (53) 0.008) 387 361 (93) 0.007

) 256 176 (69) 0.1

9) 385 14 (3.6) 0.5

15 277 24 � 14 �0.000115 277 50 � 17 0.0046.7 167 13 � 5.7 0.117 221 44 � 21 0.712 205 29 � 14 0.3

) 196 67 (34) 0.8) 196 36 (18) 0.5) 196 19 (9.7) 0.6) 196 74 (38) 0.5

15 371 36 � 15 .511 387 169 � 11 .9) 382 190 (50) 1.0) 386 336 (87) .4

1.2 369 5.8 � 1.2 1.0

) 387 157 (41) .2

) 387 167 (43) .53.9 387 10 � 4.3 .00975 326 220 � 68 .04

right cold ischemic time is used.

; IPF � idiopathic pulmonary fibrosis; PA � pulmonary artery;

atien

ients82)

or M

52 �

70 �

30 (3769 (84

37 (59

4 (4.

44 �

44 �

11 �

43 �

27 �

13 (339 (235 (13

13 (33

37 �

70 �

41 (5074 (90

.8 �

40 (49

32 (3911 �

40 �

and

ad somewhat worse survival than patients undergoing

tmutttoulo

cstvnipcp

TP

C

R

D

S

a

l

F city;r

1124 MASON ET AL Ann Thorac SurgLUNG TRANSPLANTATION FOR IPF 2007;84:1121–8

GEN

ERA

LT

HO

RA

CIC

ransplantation for other indications, when matched onultiple variables. The reasons for this difference are

nclear. It does not appear to be related to size of theransplanted organ. There was no difference in donorotal lung capacity in patients undergoing transplanta-ion for IPF and non-IPF indications. This surprising lackf size difference is attributable to our sizing strategy thatses donor organs whose predicted total lung capacity

ies between the actual and predicted total lung capacity

able 2. Recipient, Donor, and Procedure Characteristics of Pulmonary Fibrosis

haracteristic

MatchedSingle LTx

(n � 10)

naNo. (%) or

Mean � SD

ecipientDemography

Age (y)b 10 52 � 8.7Height (cm)b 10 171 � 10Women 10 5 (50)Caucasian 10 7 (70)

ComorbiditiesHistory of smoking 8 6 (75)

ImmunologyPRA �10% 10 1 (10)

Cardiac and pulmonary measuresFEV1 (% of predicted)b 8 46 � 11FVC (% of predicted) 8 42 � 17Wedge pressure (mm Hg)b 8 10 � 4.4PA systolic pressure (mm Hg) 9 45 � 23PA mean pressure (mm Hg) 9 29 � 14

Cause of posttransplantation deathRespiratory infection or failure 6 3 (50)Sepsis or multisystem organ

failure6 1 (17)

Malignancy 6 0 (0)Other 6 2 (33)

onorDemography

Age (y) 8 41 � 13Height (cm) 10 170 � 9.9Womenb 10 6 (60)Caucasian 10 10 (100)

Lung sizeTotal lung capacity (L) 8 6.2 � 1.1

Cause of deathHead trauma 10 5 (50)

urgical procedureYears from 1/1/1990 to transplantb 10 12 � 4.6Cold ischemic time (min)c 9 221 � 38

Data available. b Variable used in propensity matching of single to deft and right cold ischemic time is used.

EV1 � forced 1-second expiratory volume; FVC � forced vital capaeactive antibody; SD � standard deviation.

f the recipient. This “oversizes” donor lungs for IPF s

hests and “undersizes” lungs for patients with emphy-ema. However, the difference in survival may be relatedo the smaller chest size of IPF patients. Shrinking lungolume may have caused irreversible damage to pulmo-ary mechanics by contracting the chest wall (remodel-

ng). Matching patients with different disease states byretransplant pulmonary function tests is unfeasible be-ause each disease has its own characteristic alteration inulmonary function. For example, patients with IPF have

ts Undergoing Transplantation for Idiopathic

MatchedDouble LTx

(n � 10)

UnmatchedSingle LTx

(n � 40)

UnmatchedDouble LTx

(n � 22)

No. (%) orMean � SD na

No. (%) orMean � SD na

No. (%) orMean � SD

48 � 11 40 59 � 5.5 22 41 � 9.5174 � 9.1 40 169 � 9.1 22 172 � 11

2 (20) 40 13 (33) 22 10 (45)9 (90) 40 37 (93) 22 16 (73)

7 (78) 30 18 (60) 16 6 (38)

0 (0) 40 2 (5) 22 1 (4.5)

48 � 16 33 45 � 17 13 40 � 1344 � 16 33 47 � 14 13 36 � 1611 � 5.4 35 11 � 5.6 12 14 � 1134 � 9.9 36 41 � 10 16 50 � 2421 � 8.7 35 27 � 7.9 15 31 � 18

1 (25) 21 6 (29) 9 3 (33)1 (25) 21 6 (29) 9 1 (11)

0 (0) 21 5 (24) 9 0 (0)2 (50) 21 4 (19) 9 5 (56)

34 � 14 40 36 � 16 22 39 � 13172 � 11 40 168 � 12 22 171 � 9.3

3 (30) 40 18 (45) 22 14 (64)10 (100) 40 36 (90) 22 18 (82)

6.0 � 1.3 40 5.7 � 1.2 22 5.9 � 1.0

4 (40) 40 21 (53) 22 10 (45)

10 � 3.6 40 12 � 3.7 22 11 � 4.4285 � 64 32 201 � 54 18 292 � 84

lung transplant patients. c For double lung transplants, maximum of

LTx � lung transplant; PA � pulmonary artery; PRA � panel

atien

na

10101010

9

10

88788

44

44

10101010

10

10

1010

ouble

maller-than-predicted total lung capacity, and emphy-

spno

[ahaamesbtIls

prftpi

R

I

metrbot

t

FfiIdsml(ls

Ffiilaii

TT

R

E

L

a

vapw

1125Ann Thorac Surg MASON ET AL2007;84:1121–8 LUNG TRANSPLANTATION FOR IPF

GEN

ERA

LT

HO

RA

CIC

ema patients have larger-than-predicted total lung ca-acity; IPF patients have a restrictive pattern of pulmo-ary disease, and patients with emphysema have anbstructive pattern [16].By definition, the underlying cause of IPF is unknown

17]. Antiinflammatory treatments such as corticosteroidsnd azathioprine have been shown to be useful [9, 18];owever, disease progression usually occurs. Thesegents are part of essential immunosuppressive therapyfter lung transplantation. Idiopathic pulmonary fibrosisay carry a progressive course that is inhibited, but not

liminated, by transplantation and its immunosuppres-ion regimen. Our data suggest that the driving forceehind progression of pulmonary dysfunction may con-

inue to affect the patient even after transplantation.ndirect evidence for this exists in patients with singleung transplantation for IPF: radiographic evidence has

ig 1. Death after lung transplantation for idiopathic pulmonarybrosis (IPF; red symbols and lines) versus other indications (non-PF; blue symbols and lines). (A) Survival. Each symbol represents aeath, vertical bars are 68% confidence limits equivalent to � 1tandard error, and numbers in parentheses indicate patients re-aining at risk. Solid lines enclosed within dashed 68% confidence

imits represent parametric estimates. (B) Instantaneous risk of deathhazard function). Solid lines enclosed within dashed confidenceimits represent parametric estimates. Note expanded horizontalcale.

hown that the native lung usually demonstrates disease S

rogression [19, 20]. Nevertheless, at this point, theeason for a survival disadvantage after transplantationor IPF remains elusive and warrants focused investiga-ion. Causes of death in our patients undergoing trans-lantation for IPF were no different from those undergo-

ng transplantation for other indications.

ISK FACTORS FOR DEATH AFTER LUNG TRANSPLANTATION FOR

DIOPATHIC PULMONARY FIBROSIS. Factors associated withortality were earlier date of transplantation for IPF,

levated wedge pressure, and single lung transplanta-ion. Earlier date of transplantation as a risk factoreflects improvement in critical care management, anti-iotic prophylaxis, immunosuppressive regimens, andur own experience in caring for patients after lungransplantation that has occurred over time.

Another risk factor identified for death after transplan-ation for IPF was higher cardiac wedge pressure. This

ig 2. Survival after lung transplantation in idiopathic pulmonarybrosis (IPF; red symbols and lines) patients and patients undergo-

ng transplantation for other indications (non-IPF; blue symbols andines). Solid lines are propensity-matched patients, and dotted linesre unmatched patients. Vertical bars represent 68% confidence lim-ts equivalent to � 1 standard error, and numbers in parenthesesndicate patients remaining at risk.

able 3. Incremental Risk Factors for Death Afterransplantation for Idiopathic Pulmonary Fibrosis

isk Factor Coefficient � SE p Value Bootstrap %a

arly hazard phaseEarlier date of

transplantb�0.13 � 0.070 0.07 87

ate hazard phaseHigher wedge

pressure0.092 � 0.031 0.003 85

Single lungtransplant

1.1 � 0.50 0.03c 52

Percent of models in which variables appear; by the median rule,ariables appearing in 50% or more analyses are considered to be reliablyssociated with death with p � 0.05. b Years from 1/1/1990 to trans-lantation. c After propensity adjustment, single lung transplantationas no longer statistically significant (p � 0.3).

E � standard error.

fipmwIaeiwcashlptclf

S

P

tafptcepf

ItlptOpctmrpt

tvtdenteemld

Fft

Fpag

Fspu

1126 MASON ET AL Ann Thorac SurgLUNG TRANSPLANTATION FOR IPF 2007;84:1121–8

GEN

ERA

LT

HO

RA

CIC

nding was independent of elevated pulmonary arteryressures, which have been suggested as a risk factor forortality after transplantation for IPF [5]. Elevatededge pressures as a risk factor for transplantation for

PF may represent the effects of cardiac remodeling. Kroftnd colleagues [21] used magnetic resonance imaging toxamine cardiac function in patients with IPF and foundmpaired right ventricular diastolic function comparedith control subjects. These patients also had lower

ardiac output, although the study population was smallnd numbers did not reach statistical significance. Recenttudies in patients with pulmonary arterial hypertensionave shown that in the presence of a low cardiac index,

ower, not higher, pulmonary pressures portend a worserognosis [22–24]. A higher wedge pressure in our pa-

ients may reflect compensation occurring to maintainardiac output. Finally, single lung transplantation was ateast a marker for higher mortality after transplantationor IPF.

ig 3. Percent of patients undergoing double lung transplantationor idiopathic pulmonary fibrosis. Numbers above bars represent pa-ients undergoing transplantation.

ig 4. Percent of patients undergoing lung transplantation for idio-athic pulmonary fibrosis who received double versus single lungsccording to recipient age at transplantation. Dots indicate actual

erouped percentage; solid line is trend line.

INGLE VERSUS DOUBLE LUNG TRANSPLANTATION FOR IDIOPATHIC

ULMONARY FIBROSIS. Choice of single versus double lungransplantation in nonsuppurative lung disease remains

hotly debated topic. However, most of the debate hasocused on transplantation for emphysema. Results ap-ear to support a survival advantage for double lung

ransplantation, although this is confounded by in-reased risk with older recipient age and higher periop-rative mortality [7, 8, 25–27]. Despite this, most trans-lant physicians advocate double lung transplantation

or younger patients with emphysema.Choice of single versus double lung transplantation for

PF is far less clear. One study comparing the twoechniques demonstrated a survival advantage for singleung transplantation in patients with IPF [6]. However,erioperative mortality was higher for double lung

ransplantation, and follow-up was limited to 3 years.ne-month mortality was 10% for single lung trans-lantation and 21% for double lung transplantation,ompared with 6% for both single and double lungransplantation in our study. Increased perioperative

ortality for double lung transplantation in that expe-ience may have negated any survival benefit, withotential benefits not realized even by intermediate-

erm follow-up.Another study comparing single versus double lung

ransplantation for IPF failed to demonstrate any sur-ival advantage of double lung transplantation, al-hough the study was small, with 32 single lung and 13ouble lung transplantations [28]. The largest studyvaluating lung transplantation for IPF used the Inter-ational Society for Heart and Lung Transplant Regis-

ry for a cohort study of 830 patients, focusing on theffect of preoperative pulmonary artery pressure [5]. Itxcluded 667 patients because of incomplete data, wasulti-institutional, had high 90-day mortality, and

imited follow-up to 90 days. Results showed thatouble lung transplantation carried greater risk for

ig 5. Survival after transplantation for idiopathic pulmonary fibro-is according to single (red lines) or double (blue lines) lung trans-lantation. Solid lines are 10 matched pairs, and dashed lines arenmatched patients. Format is as in Figure 2.

arly mortality, and elevated pulmonary artery pres-

spo

ttlvttrstmstifddsrfimti

ugwfwdMad

LTcotgtwepOopls

CSasltw

ddsowitsTprd

Ws

R

1

11

1

1

1

1

1127Ann Thorac Surg MASON ET AL2007;84:1121–8 LUNG TRANSPLANTATION FOR IPF

GEN

ERA

LT

HO

RA

CIC

ures increased risk of death after single lung trans-lantation. The authors were unable to recommendne procedure over the other.In our study, patients receiving double lung transplan-

ation for IPF had better risk-unadjusted survival thanhose receiving a single lung. In fact, survival after doubleung transplantation for IPF closely approximated sur-ival for double lung transplantation for non-IPF indica-ions. Reasons for improved survival with double lungransplantation are likely related to enhanced pulmonaryeserve and improved respiratory mechanics. Improvedurvival with double lung transplantation for IPF was notrue for matched patients. However, we were able to

atch only 10 pairs of sicker patients in our analysis. Thismall number of matched patients reflects a strong selec-ion bias that was practiced at our program when choos-ng patients for single versus double lung transplantationor IPF and makes comparison difficult. The strongestiscriminator was age, with younger patients receivingouble lung transplantation and older patients receivingingle. Although advanced age has been considered as aisk factor for double lung transplantation, we did notnd it to be associated with increased mortality in ourultivariable analysis, supporting emerging literature

hat double lung transplantation can be safely performedn older patients [29].

Interestingly, survival was worse for matched thannmatched patients in both the single and double lungroups. This provides some clues about contributors toorse outcomes in patients undergoing transplantation

or IPF. Matched single lung transplantation patientsere younger and received lungs from older donors thanid unmatched single lung transplantation patients.atched double lung transplantation patients were older

nd more likely to be smokers than were unmatchedouble lung transplantation patients.

imitationshe primary limitation of this study is that it represents alinical experience at a single center with a small numberf patients undergoing transplantation for IPF. In addi-ion, although we were able to show that patients under-oing transplantation for IPF had worse outcomes thanhose undergoing transplantation for other indications,e were not able to identify the reasons for this differ-

nce. Finally, choice of single versus double lung trans-lantation in our group was influenced by selection bias.ur lung transplant committee chooses patients based

n objective as well as subjective criteria that maderopensity matching in our analysis difficult. This high-

ights the need for a prospective, randomized trial ofingle versus double lung transplantation for IPF.

onclusionsurvival after lung transplantation for IPF is worse thanfter other indications for lung transplantation. However,urvival of these patients is better if they receive twoungs and approaches survival of patients undergoingransplantation for non-IPF indications. Our numbers

ere small, however, and this advantage could not be

emonstrated in our matched analysis. Choice of proce-ure must be balanced against allocating a scarce re-ource with the potential to save two lives rather thanne. However, given the national trend of decreasingaiting times, as well as decreasing waiting list mortality

n the face of an increasing proportion of double lungransplantations being performed [4], the primary focushould be on improving individual patient outcome.herefore, we cautiously recommend double lung trans-lantation for patients with IPF, although a prospective,andomized trial will be necessary to resolve this issueefinitively.

e acknowledge and appreciate the expert editing and manu-cript preparation by Tess Parry.

eferences

1. Gross TJ, Hunninghake GW. Idiopathic pulmonary fibrosis.N Engl J Med 2001;345:517–25.

2. Thabut G, Mal H, Castier Y, et al. Survival benefit of lungtransplantation for patients with idiopathic pulmonary fi-brosis. J Thorac Cardiovasc Surg 2003;126:469–75.

3. Alalawi R, Whelan T, Bajwa RS, Hodges TN. Lung trans-plantation and interstitial lung disease. Curr Opin PulmMed 2005;11:461–6.

4. Orens JB, Shearon TH, Freudenburger RS, Conte JV,Bhorade SM, Ardehali A. Thoracic organ transplantation inthe United States, 1995–2004. Am J Transplant 2006;6:1188–97.

5. Whelan TP, Dunitz JM, Kelly RF, et al. Effect of preoperativepulmonary artery pressure on early survival after lungtransplantation for idiopathic pulmonary fibrosis. J HeartLung Transplant 2005;24:1269–74.

6. Meyer DM, Edwards LB, Torres F, Jessen ME, Novick RJ.Impact of recipient age and procedure type on survival afterlung transplantation for pulmonary fibrosis. Ann ThoracSurg 2005;79:950–8.

7. Cassivi SD, Meyers BF, Battafarano RJ, et al. Thirteen-yearexperience in lung transplantation for emphysema. AnnThorac Surg 2002;74:1663–70.

8. Bavaria JE, Kotloff R, Palevsky H, et al. Bilateral versus singlelung transplantation for chronic obstructive pulmonary dis-ease. J Thorac Cardiovasc Surg 1997;113:520–8.

9. American Thoracic Society. Idiopathic pulmonary fibrosis:diagnosis and treatment. International consensus statement.American Thoracic Society (ATS), and the European Respi-ratory Society (ERS). Am J Respir Crit Care Med 2000;161:646–64.

0. Blackstone EH, Naftel DC, Turner ME Jr. The decompositionof time-varying hazard into phases, each incorporating aseparate stream of concomitant information. J Am Stat Assoc1986;81:615–24.

1. Breiman L. Bagging predictors. Mach Learn 1996;24:123–40.2. Blackstone EH. Breaking down barriers: helpful break-

through statistical methods you need to understand better.J Thorac Cardiovasc Surg 2001;122:430–9.

3. Rosenbaum PR, Rubin DB. The central role of the propensityscore in observational studies for causal effects. Biometrika1983;70:41–55.

4. Blackstone EH. Comparing apples and oranges. J ThoracCardiovasc Surg 2002;123:8–15.

5. Mal H, Brugiere O, Dauriat G, et al. [Lung transplantation inpatients with pulmonary fibrosis]. Rev Pneumol Clin 2005;61:232–8.

6. Khalil N, O’Connor R. Idiopathic pulmonary fibrosis: cur-

rent understanding of the pathogenesis and the status oftreatment. CMAJ 2004;171:153–60.

1

1

1

2

2

2

2

2

2

2

2

2

2

A

V

R

Af

mpnvvev

prB

D

Aa

c

(

tl

I

1128 MASON ET AL Ann Thorac SurgLUNG TRANSPLANTATION FOR IPF 2007;84:1121–8

GEN

ERA

LT

HO

RA

CIC

7. Abdelaziz MM, Samman YS, Wali SO, Hamad MM. Treat-ment of idiopathic pulmonary fibrosis: is there anythingnew? Respirology 2005;10:284–9.

8. Hospenthal MA. Diagnosis and management of idiopathicpulmonary fibrosis: implications for respiratory care. RespirCare 2006;51:382–91.

9. Lok SS, Smith E, Doran HM, Sawyer R, Yonan N, Egan JJ.Idiopathic pulmonary fibrosis and cyclosporine: alesson from single-lung transplantation. Chest 1998;114:1478 – 81.

0. Wahidi MM, Ravenel J, Palmer SM, McAdams HP. Progres-sion of idiopathic pulmonary fibrosis in native lungs aftersingle lung transplantation. Chest 2002;121:2072–6.

1. Kroft LJ, Simons P, van Laar JM, de Roos A. Patients withpulmonary fibrosis: cardiac function assessed with MR im-aging. Radiology 2000;216:464–71.

2. Sandoval J, Bauerle O, Palomar A, et al. Survival in primarypulmonary hypertension. Validation of a prognostic equa-tion. Circulation 1994;89:1733–44.

3. Sitbon O, Humbert M, Nunes H, et al. Long-term intrave-nous epoprostenol infusion in primary pulmonary hyperten-sion: prognostic factors and survival. J Am Coll Cardiol2002;40:780–8.

4. McLaughlin VV, Sitbon O, Badesch DB, et al. Survival withfirst-line bosentan in patients with primary pulmonary hy-pertension. Eur Respir J 2005;25:244–9.

5. Angles R, Tenorio L, Roman A, Soler J, Rochera M, deLatorre FJ. Lung transplantation for emphysema. Lung hy-perinflation: incidence and outcome. Transpl Int 2005;17:810–4.

6. Bando K, Paradis IL, Keenan RJ, et al. Comparison ofoutcomes after single and bilateral lung transplantation forobstructive lung disease. J Heart Lung Transplant 1995;14:692–8.

7. Pochettino A, Kotloff RM, Rosengard BR, et al. Bilateralversus single lung transplantation for chronic obstructivepulmonary disease: intermediate-term results. Ann ThoracSurg 2000;70:1813–9.

8. Meyers BF, Lynch JP, Trulock EP, Guthrie T, Cooper JD,Patterson GA. Single versus bilateral lung transplantationfor idiopathic pulmonary fibrosis: a ten-year institutionalexperience. J Thorac Cardiovasc Surg 2000;120:99 –107.

9. Palmer SM, Davis RD, Simsir SA, et al. Successful bilaterallung transplant outcomes in recipients 61 years of age and

ppendix

ariables Used in Analyses

ECIPIENT Demographic: Sex,* age,�† race (Caucasian,* African-merican), height,† weight,* weight-to-height ratio, body sur-

ace area, body mass indexCardiac and pulmonary: Blood pressure (systolic,* diastolic,ean), cardiac index,* cardiac output, central venous pressure,*

ulmonary artery pressure (systolic,* diastolic, mean), pulmo-ary vascular resistance,* wedge pressure,† forced expiratoryolume in 1 second (actual† and percent of predicted), forcedital capacity (actual and percent of predicted), ratio of forcedxpiratory volume in 1 second percent of predicted to forcedital capacity percent of predicted

Comorbidities: History of hypertension,* history of smokingSerology/immunology: Blood type (A, B, AB,* O, Rh positive),

retransplantation serum creatinine,* pretransplantation panel-eactive antibody, cytomegalovirus (immunoglobulin G), Epstein-arr virus (immunoglobulin G)

ONOR Demographic: Sex,�† age,* race (Caucasian,* African-merican), height, weight,* weight-to-height ratio, body surfacerea, body mass index

Comorbidities: History of hypertension*Serology: Blood type (A,* B, O, Rh positive), serum creatinine,

ytomegalovirus (immunoglobulin G)Lung size: Total lung capacity*Cause of death: Cerebral bleed, stroke, head trauma*Donor/Recipient: Rh-factor mismatch, cytomegalovirus

CMV) mismatch (CMV� donor to CMV� recipient)Transplantation: Years from January 1, 1990, to transplanta-

ion,�† double lung transplantation,* right side transplantation,eft side transplantation,* cold ischemic time*

____________________*Indicates variables used in propensity matching IPF to non-

PF patients.†Indicates variables used in propensity matching single to

older. Transplantation 2006;81:862–5. double lung transplantation in IPF patients.