Quality-of-life and quality-adjusted survival (Q-TWiST) in patients receiving lapatinib in combination with paclitaxel as first-line treatment for metastatic breast cancer

Copyright ©

2010 Inform

a UK Limite

d

Not for S

ale or Commerc

ial Distri

bution

Unauthoriz

ed use prohibite

d. Auth

orised users

can download,

display, view and print a

single copy for p

ersonal u

se

Current Medical Research & Opinion Vol. 26, No. 4, 2010, 767–775

0300-7995 Article 5249/459595

doi:10.1185/03007991003590860 All rights reserved: reproduction in whole or part not permitted

Original articleQuality-of-life and quality-adjusted survival(Q-TWiST) in patients receiving lapatinib incombination with paclitaxel as first-linetreatment for metastatic breast cancer

Beth SherrillRTI Health Solutions, Research Triangle Park, NC, USA

Angelo Di Leo“Sandro Pitigliani” Medical Oncology Unit, Prato, Italy

Mayur M. AmonkarGSK Oncology, Collegeville, PA, USA

Yun WuMStat, RTI Health Solutions, Research Triangle Park,

NC, USA

Zanete ZvirbuleRiga Eastern University Hospital Latvian Oncology

Centre, Rıga, Latvia

Zeba AzizAllama Iqbal Medical College, Lahore, Pakistan

Jose BinesNational Cancer Institute, Rio De Janeiro, Brazil

Henry L. GomezInstituto Nacional de Enfermedades Neoplasicas,

Lima, Peru

Address for correspondence:Beth Sherrill, 3040 Cornwallis Road, PO Box 12194,

Research Triangle Park, NC 27709-2194, USA.

Tel.: þ1 (919) 541-1233; Fax: þ1 (919) 541-7222;

[email protected]

Key words:Cancer, metastatic breast – ErbB2þ – HER2 –

Lapatinib – Paclitaxel, Q-TWiST – Quality-of-life

Accepted: 4 January 2010; published online: 22 January 2010

Citation: Curr Med Res Opin 2010; 26:767–75

Abstract

Background:

In a phase 3 randomized, multicenter, double-blind, placebo-controlled study, first-line therapy with

lapatinib plus paclitaxel significantly improved clinical outcomes based on a pre-planned analysis of

ErbB2þ metastatic breast cancer patients (GSK Study #EGF30001; ClinicalTrials.gov identifier:

NCT00075270). Patients with ErbB2� or untested did not significantly benefit. This article focuses on

the quality of life (QOL) and quality-adjusted survival outcomes (Q-TWiST) in the study.

Methods:

QOL was assessed using the Functional Assessment of Cancer Therapy–Breast (FACT-B). Changes from

baseline were analyzed using ANCOVAs, repeated measures and pattern mixture modeling. The Q-TWiST

method was used to examine the trade-off between toxicities and delayed progression.

Results:

The study included 579 subjects, of whom 86 were ErbB2þ. In the ITT population, no significant differences

in QOL or Q-TWiST scores were observed. In the ErbB2þ subgroup, the lapatinib plus paclitaxel (Lþ P) arm

demonstrated stable FACT-B scores over the first year, while average scores for patients on Pþ placebo

(Pþ pla) monotherapy decreased (change from baseline: Lþ P, p¼ 0.99; Pþ pla, p¼ 0.01). Clinically

meaningful differences were observed between treatment arms on the FACT-B, Trial Outcome Index and

breast cancer subscale scores. Pattern mixture models suggested more QOL differentiation between

treatments among patients who progressed or withdrew early. Q-TWiST differences between the arms in

the ErbB2þ subgroup ranged from 2 to 15 weeks with an Lþ P advantage across all utility weight

combinations.

Conclusions:

In the ITT population, results provide no evidence of QOL differences between treatment groups. In a small,

prospectively-defined subgroup of ErbB2þ patients, Lþ P resulted in more stable QOL and more quality-

adjusted survival than paclitaxel monotherapy, representing clinically important differences between

treatments.

Introduction

Breast cancer is the most common malignancy and the second most commoncause of cancer-related death in Western European and North Americanwomen. The incidence rate of breast cancer among developed nations(68/100 000) is much higher than the worldwide rate (37/100 000)1.

! 2010 Informa UK Ltd www.cmrojournal.com QOL and Q-TWiST for lapatinibþ paclitaxel in metastatic breast cancer Sherrill et al. 767

Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

There were approximately 430 000 breast cancer casesdiagnosed in Europe in 20062 and 250 000 in the US in20083. In the subgroup of breast cancer patients who over-express the ErbB2 (Her2/neu) oncogene, which comprisesapproximately 25% of all human breast cancers, the dis-ease is particularly virulent and is associated with a greaterrisk for disease progression and death4.

Women with metastatic breast cancer often undergoseveral rounds of treatment that may include hormonetherapy, radiation, and chemotherapy. Furthermore, astheir disease progresses, they often experience painfuland debilitating metastases to the brain, bones, andother organs, all of which can impact quality of life(QOL)5–8. Hence, in addition to showing clinical benefit,an important goal of treating women with metastaticbreast cancer is to provide palliation and improve or main-tain QOL.

The current standard first line of treatment for meta-static ErbB2þ breast cancer is trastuzumab combined witha taxane9. This combination has been shown to signifi-cantly prolong response duration, time to disease progres-sion and overall survival compared with chemotherapyalone for patients who have not received prior therapyfor metastatic disease10,11. However, most patients developresistance to trastuzumab, and their disease progresseswithin 1 year12,13.

Lapatinib (Tykerb/Tyverb; GlaxoSmithKline [GSK]) isan epidermal growth factor (EGFR) inhibitor currentlyapproved in the US for use in combination with capecita-bine (Xeloda, Roche) for the treatment of advanced ormetastatic ErbB2-positive breast cancer in women whohave received prior therapy, including trastuzumab. Thepositive effect of lapatinib on quality of life in this settinghas previously been shown14,15. Lapatinib has also beenstudied in combination with trastuzumab in the ErbB2þpopulation, with results showing prolonged progression-free survival (PFS) compared with trastuzumab alone16,without any denigration of quality of life17. It has alsobeen evaluated in combination with letrozole as first-linetherapy in hormone receptor-positive metastatic breastcancer patients with results showing significantly longerprogression-free survival versus letrozole alone in theErbB2þ and the ITT populations18.

In a trial of lapatinib with paclitaxel (LþP) or pacli-taxel with placebo (Pþpla) as first-line treatment forErbB2-negative (or unknown) metastatic breast cancer,the overall study population did not show a significantdifference in time to disease progression (TTP) betweengroups but did show a better overall response rate for thecombination group than the paclitaxel group. Archivedtumor specimens were available for central testing ofErbB2 status and were used to identify a subset of patientswho were ErbB2þ based on documented amplification ofErbB2 by fluorescence in situ hybridization (FISH) orimmunohistochemistry (IHC3þ). The ErbB2þ patients

taking combination Lþ P had a statistically significantadvantage over the group on paclitaxel monotherapy inTTP, and median survival was 22.2 weeks longer(p¼ 0.36)19. This article focuses on the quality of life(QOL) and quality-adjusted survival (Q-TWiST) out-comes for the ITT population and for the prospectivelydefined subgroup of confirmed ErbB2þ patients in thestudy.

Methods

Clinical trial

Analyses were performed using data from a phase 3 rando-mized, placebo-controlled, double-blind, multicenterstudy (GSK Study #EGF30001; ClinicalTrials.gov identi-fier: NCT00075270) that compared oral Lþ P withPþ pla in women previously untreated for advanced ormetastatic breast cancer that was negative or untestedfor ErbB2 overexpression19.

Subjects were randomized to receive either:� Oral lapatinib (1500 mg daily) plus paclitaxel

(175 mg/m2 intravenously [IV] over 3 hours every 3weeks); or

� Paclitaxel (175 mg/m2 IV over 3 hours every 3 weeks)plus placebo.

Treatment was administered until disease progression orwithdrawal. The primary endpoint was time to tumor pro-gression (TTP), defined as the interval between random-ization and the date of disease progression or death due tobreast cancer. Progression was determined by investigatorsaccording to RECIST (response evaluation criteria in solidtumors). Efficacy assessments were performed at 9 weeks,then every 12 weeks, and at the end of treatment. If newanti-cancer therapy was started prior to or in the absence ofdisease progression, TTP was censored at the date of thelast radiological assessment prior to start of alternate ther-apy. After treatment ended, subjects were followed for sur-vival at approximately 12-week intervals. Further details ofstudy methods and efficacy/safety results have been pre-viously reported19. Prior to break of the study blind, centraltesting of tissue samples was undertaken and the ErbB2þpopulation was identified as a preplanned subset foranalyses.

The analysis presented here uses data from theOctober 2006 lock date and survival data throughFebruary 2007.

Quality of life

QOL was assessed using the Functional Assessment ofCancer Therapy–Breast (FACT-B) questionnaire(Version 4, 1997). The FACT-B includes items specificfor women with breast cancer, such as questions about

Current Medical Research & Opinion Volume 26, Number 4 April 2010

768 QOL and Q-TWiST for lapatinibþ paclitaxel in metastatic breast cancer Sherrill et al. www.cmrojournal.com ! 2010 Informa UK Ltd

Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

the effect of weight and hair loss, in addition to items forcancer patients in general. Five dimensions (physicalwell-being, social/family well-being, emotional well-being,functional well-being, and breast cancer subscale) are sum-marized into a FACT-B score and three dimensions (phys-ical well-being, functional well-being, and breast cancersubscales) are summed to form the trial outcome index(TOI)20. High scores indicate better QOL. In a previousstudy, a clinically meaningful change was estimated to be7–8 points for the FACT-B total score, 5–6 points for theTOI, and 2–3 points for the breast cancer subscale(BCS)21.

Patients completed a FACT-B questionnaire at screen-ing, at 9 weeks, thereafter at intervals of 12 weeks, and atdiscontinuation of randomized therapy. Changes frombaseline in the QOL scores were analyzed by parametricanalysis of covariance at each time point, using baseline asa covariate. Based on results, additional post hoc analyseswere performed on the ErbB2þ subgroup using models thataccount for intra-patient correlations. Repeated measuresanalyses were conducted on the FACT-B, TOI, and breastcancer subscale using linear models with an autoregressivecovariance structure. The QOL change from baseline overthe first year was modeled, adjusted for baseline value, withscores from discontinuation assessments assigned to thenext scheduled visit.

The repeated measures model accommodates data thatare missing at random, but does not account for nonran-dom missingness. Because the treatment period for anysubject was dependent on treatment response and QOLwas not assessed after treatment ended, the duration offollow-up varied among subjects and may have beentreatment-related. Therefore, a pattern mixture modelwas run as a sensitivity analysis to assess the impact ofthe missing data configuration on results. Each treatmentgroup was stratified into patients who progressed or with-drew from treatment within 6 months and those whostayed on treatment longer. It was not possible to furtherelaborate the missing data configuration due to the smallsample size.

Quality-adjusted survival

Since QOL data were not collected after progression, thequality-adjusted time without symptoms or toxicity(Q-TWiST) method was applied to evaluate the overallperiod from start of therapy until death. This methodincorporates the trade-off between treatment efficacyand safety, providing a single measure for comparingquality-adjusted survival between groups. The Q-TWiSTapproach partitions survival time into three health states:TOX (toxicity of treatment); TWiST (time without symp-toms or toxicity); and REL (time from relapse untildeath)22.

Any day with Grade 3/4 adverse events (AEs) afterrandomization and prior to disease progression is includedin the TOX state. As per convention, AEs occurring afterrelapse were not included in the TOX state. The TWiSTperiod extends from the end of the cumulative AE perioduntil disease progression, censoring for progression ordeath. For each treatment group, TWiST duration isderived as the difference between the average progres-sion-free survival time and the average time with toxici-ties. Thus, it represents the relatively healthier periodduring which the patient is experiencing neither symptomsof progression nor treatment toxicities. The REL state isderived as the difference between overall survival andprogression-free survival. Since all patients in this studyhad advanced or metastatic disease, the REL state hererefers to the period after further disease progression.

Q-TWiST was calculated by applying a range of utilityvalues to the mean time in each health state by treatmentgroup as:

Q-TWiSTi ¼ ðuTOX�TOXiÞþTWiSTiþðuREL�RELiÞ

where TOX, TWiST, and REL represent the mean healthstate durations based on the product-limit method. Theparameters uTOX and uREL denote utility weights rangingfrom 0 to 1. These hypothetical utility weights wereapplied to time in the TOX and REL health states to reflectthe relatively poorer utility of these states compared totime with better health during TWiST. Q-TWiST differ-ences between treatment groups were calculated for eachcombination of utility weights to determine the thresholdutility values at which differences were statisticallysignificant.

Results

ITT population

In the ITT population (N¼ 579), patient characteristicswere balanced for baseline characteristics including age,Eastern Cooperative Oncology Group (ECOG) status,stage of disease and number of metastatic sites19. On aver-age, subjects in the treatment arms had similar baselineQOL subscale and total scores (Table 1).

QOL scores were generally stable over time in bothgroups up to the point of discontinuation of treatment(Table 2). The breast cancer subscale score increased inthe Lþ P arm after week 9, whereas it decreased over timein the paclitaxel monotherapy arm. However, neitherreached a clinically meaningful change (2–3 points).Other scales demonstrated a similar pattern. Changesfrom baseline in observed FACT-B and TOI scores didnot reach MID levels during treatment, except for declinesat the withdrawal visit in the Pþ pla group. At the with-drawal visit, average scores were smaller than at baseline



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

for the FACT-B total and TOI scores in both arms. Whenanalyzed for change from baseline, point estimates forthe Lþ P group were slightly more favorable than in thePþ pla group, although none of the differences were sta-tistically or clinically significant.

To evaluate the entire patient experience includingtime beyond the treatment period, quality-adjusted sur-vival was assessed over the median follow-up period of96 weeks. Q-TWiST differences between treatmentgroups were not statistically significant for any combina-tion of utility weights in the ITT population, although alldifferences favored the Lþ P arm (Table 3). When a day oftoxicity or postprogression was counted at a utility equal to

2 days of TWiST, the Q-TWiST difference was 2.5 weeksin favor of Lþ P, although not statistically significant. In asensitivity analysis with TOX expanded to include all AEs,some combinations of utility weights resulted in Q-TWiSTdifferences favoring the combination group (LþP), butthe magnitude of difference was 1 month or less.

ErbB2þ population

The baseline characteristics of patients who were identi-fied as ErbB2þ (Lþ P, n¼ 49; Pþ pla, n¼ 37) werewell-balanced across arms with regard to age, ECOGstatus, and prior anthracycline use. More patients in the

Table 2. Summary of quality of life score changes from baseline by visit in a study of first-line therapy with lapatinib–paclitaxel.

ITT population ErbB2þ subgroup

Lapatinibþ paclitaxel Paclitaxelþ placebo Lapatinibþ paclitaxel Paclitaxelþ placebo

n Mean (SD) n Mean (SD) n Mean (SD) n Mean (SD)

Breast cancer subscaleWeek 9 211 �0.4 (4.85) 235 �0.7 (5.04) 42 1.5 (4.80) 32 �1.8 (5.51)Week 21 126 0.6 (4.88) 127 �0.9 (4.75) 34 0.8 (4.43) 19 �2.0 (3.56)Week 33 73 0.4 (4.77) 66 �0.3 (4.43) 23 1.1 (4.34) 10 �0.7 (5.77)Week 45 38 1.4 (6.35) 39 �0.2 (3.74) 14 3.0 (6.09) 3 �0.7 (0.62)Withdrawal 194 �0.6 (4.98) 217 �1.0 (5.25) 34 0.7 (7.33) 29 �3.2 (5.33)

FACT-B total*Week 9 208 �1.1 (16.09) 230 �2.3 (16.27) 41 1.6 (17.57) 32 �4.3 (20.78)Week 21 126 1.0 (16.63) 125 �1.3 (17.13) 34 �0.3 (15.85) 19 �1.3 (16.59)Week 33 72 1.4 (17.39) 64 �2.0 (12.75) 22 0.4 (16.09) 10 �1.9 (11.12)Week 45 35 2.3 (22.05) 38 �1.4 (10.79) 14 6.9 (19.45) 3 2.1 (5.40)Withdrawal 190 �5.0 (19.53) 212 �8.4 (17.99) 33 1.5 (22.74) 29 �11.6 (16.94)

TOIyWeek 9 208 �2.3 (12.32) 232 �2.6 (11.88) 41 0.9 (13.48) 32 �3.5 (13.63)Week 21 126 �0.5 (11.99) 125 �2.7 (12.42) 34 �0.4 (11.96) 19 �2.5 (11.44)Week 33 72 �0.1 (12.83) 65 �2.9 (9.05) 22 0 (9.97) 10 �1.1 (9.56)Week 45 36 1.5 (15.91) 38 �2.8 (8.98) 14 5.4 (11.73) 3 0 (5.16)Withdrawal 190 �4.4 (14.30) 212 �6.1 (12.84) 33 2.0 (16.30) 29 �7.9 (11.14)

n, number of subjects whose overall item response rate was greater than 80% on score.*Functional Assessment of Cancer Therapy–Breast (FACT-B) total score is the sum of the five subscale scores.yTrial outcome index (TOI) is the sum of the physical well-being, functional well-being, and breast cancer subscale scores.Higher scores indicate better QOL. Clinically meaningful differences: 7–8 points for FACT-B; 5–6 points for TOI; and 2–3 points for BCS.

Table 1. Baseline quality of life subscales and total scores in a study of first-line therapy with lapatinib–paclitaxel.

ITT population ErbB2þ subgroup

Lapatinibþ paclitaxel Paclitaxelþ placebo Lapatinibþ paclitaxely Paclitaxelþ placebo

n Mean (SD) n Mean (SD) n Mean (SD) n Mean (SD)

Physical well-being (max 28) 286 20.4 (5.96) 284 20.7 (5.41) 48 19.2 (6.63) 37 20.5 (5.84)Social/family well-being (max 28) 286 20.5 (5.62) 283 20.7 (5.63) 48 19.1 (5.82) 37 19.3 (5.64)Emotional well-being (max 24) 285 15.0 (4.63) 283 15.6 (4.85) 48 15.7 (4.32) 37 16.6 (4.17)Functional well-being (max 28) 285 16.3 (6.05) 282 16.9 (5.58) 48 15.8 (6.49) 37 17.6 (5.36)Breast cancer subscale (max 36) 282 19.2 (5.28) 283 19.6 (5.87) 48 21.1 (5.30) 37 23.4 (6.06)FACT-B total score (max 144) 281 94.6 (19.07) 278 97.0 (18.82) 48 90.8 (19.67) 37 97.3 (18.74)FACT–G score (max 108) 283 72.1 (16.20) 279 74.1 (15.07) 48 69.8 (16.59) 37 74.0 (14.88)TOI (max 92) 281 59.2 (13.62) 279 60.6 (13.37) 48 56.1 (15.60) 37 61.5 (13.28)

FACT-B, Functional Assessment of Cancer Therapy-Breast; FACT–G, FACT-General; TOI, Trial Outcome Index.yOne patient in the Lþ P group did not complete a baseline assessment, and two patients (one in each group) did not complete any post-baseline assessments.These patients were excluded from QOL models.Higher scores indicate better QOL.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

Lþ P group than in the Pþ pla group had visceral disease(69 vs. 51%) and more were at stage IV (88 vs. 78%)19.The length of paclitaxel exposure during the treatmentperiod was similar between arms, typically ending after 6q3week cycles (median exposure: LþP¼ 15.1 weeks;Pþ pla¼ 16.1 weeks). The smaller average baselineFACT-B, FACT-G, and TOI scores for patients receivingLþ P are close to clinically meaningful differences, sug-gesting that in the ErbB2þ subgroup this arm had lowerQOL before beginning treatment than patients whoreceived paclitaxel alone (Table 1).

Observed QOL scores indicate that the LþP armremained generally stable after baseline while the pacli-taxel arm declined on average (Table 2). Only the differ-ences at withdrawal were significantly different betweenarms based on tests at separate time points (FACT-Bp¼ 0.04; FACT-G p¼ 0.07; TOI p¼ 0.03). This findingled to additional analyses on the ErbB2þ subgroup thatincorporated QOL scores across the multiple time points,including treatment discontinuation assessments assignedto the next scheduled visit. The repeated measures modelswere restricted to 1 year, since few patients remained onstudy after that time. As seen in Figure 1(a), these analysesconfirmed that the Lþ P group had stable FACT-B scoresafter baseline (p¼ 0.99 change from baseline). The scoresfor the Pþ pla group declined 7.7 points on average(p¼ 0.01 change from baseline), reaching a clinically

meaningful decrease in QOL within 21 weeks. The averagedifference between groups (p¼ 0.05) varied over time(p¼ 0.07 Trt*Time interaction). Estimated treatment dif-ferences at 9, 21, 33, and 45 weeks are shown in Table 4.Similar results are seen for the TOI, which focuses on thephysical aspects of QOL (Figure 1(b)). Again, the Lþ Pgroup had stable QOL on average after baseline (p¼ 0.82),while the Pþ pla group had statistically and clinically sig-nificant declines from the second assessment onward(p¼ 0.01). The average difference between treatmentgroups on the TOI was statistically significant (p¼ 0.03).On the BCS (Figure 1(c)), the decline in the Pþ pla groupreached a clinically meaningful decline of 2–3 points onlyat the second assessment. The overall difference betweengroups was significant (p50.01).

For the sensitivity analyses, each treatment group wasdivided into patients who progressed or withdrew fromtreatment within 6 months (Lþ P, n¼ 18; Pþ pla,n¼ 23) and those who stayed on treatment longer(Lþ P, n¼ 30; Pþ pla, n¼ 14). This grouping capturedthe missing data patterns, since QOL data were not col-lected after treatment ended and all missing data were atthe end of subjects’ records. In the pattern mixture models,the three-way interaction of treatment effect with assess-ment time and dropout pattern measures the impact ofmissing data patterns on the treatment effects. This inter-action was not statistically significant for any of the QOL

Table 3. Q-TWiST differences (weeks) between treatment arms in a study of first-line therapy with lapatinib–paclitaxel.

Utility per phase ITT population ErbB2þ subgroup

Tox TWiST Relapse Difference inQ-TWiST (weeks)*

p-value Difference inQ-TWiST (weeks)*

p-value

0 1 0 0.36 0.8587 7.58 0.14070.25 1 0 0.82 0.6691 9.46 0.04920.5 1 0 1.25 0.5290 11.34 0.01340.75 1 0 1.52 0.4884 13.22 0.00331 1 0 1.72 0.4952 15.10 0.00090 1 0.25 0.95 0.6287 6.15 0.22060.25 1 0.25 1.41 0.4479 8.03 0.08440.5 1 0.25 1.85 0.3349 9.91 0.02430.75 1 0.25 2.12 0.3175 11.80 0.00581 1 0.25 2.31 0.3439 13.68 0.00150 1 0.5 1.56 0.4317 4.73 0.37660.25 1 0.5 2.02 0.2779 6.61 0.18450.5 1 0.5 2.45 0.1964 8.49 0.07200.75 1 0.5 2.71 0.1935 10.37 0.02361 1 0.5 2.90 0.2271 12.25 0.00740 1 0.75 2.19 0.2880 3.30 0.58490.25 1 0.75 2.64 0.1685 5.18 0.36330.5 1 0.75 3.06 0.1132 7.06 0.19520.75 1 0.75 3.31 0.1161 8.94 0.09221 1 0.75 3.48 0.1469 10.82 0.04060 1 1 2.83 0.1948 1.87 0.78920.25 1 1 3.28 0.1058 3.75 0.57480.5 1 1 3.68 0.0686 5.63 0.38330.75 1 1 3.90 0.0725 7.51 0.23481 1 1 4.07 0.0965 9.39 0.1351

*Positive values indicate more quality-adjusted time for the lapatinibþ paclitaxel group; Q-TWiST, quality-adjusted time without symptoms ortoxicity; TWiST, time without symptoms or toxicity.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

scores (FACT-B, p¼ 0.40; TOI, p¼ 0.32; BCS, p¼ 0.89).However, results suggest a differential effect of treatmentby dropout status, with greater difference between treat-ment arms occurring in the group who ended treatmentwithin 6 months of baseline; most of these patients

progressed or died (Lþ P, 78%; Pþ pla, 91%). The smallerdifference in QOL scores between treatments among thosewho stayed on study through the week 33 or week 48 visitwas at least partially due to patients who remained stablefor a long period and still had not progressed after 1 year

3.0000

(a) (b)

(c)

0

–3.0000

–6.0000

–9.0000

–12.0000

Treatment

P + placebo

0 10 20 30Weeks

40 50

Est

imat

ed m

ean

chan

ge fr

om b

asel

ine

for

FAC

T-B

3.0000

0

–3.0000

–6.0000

–9.0000

–12.0000

Treatment

P + placebo

0 10 20 30Weeks

40 50

Est

imat

ed m

ean

chan

ge fr

om b

asel

ine

for T

OI

3.0000

0

–3.0000

–6.0000

–9.0000

–12.0000

Treatment

P + placebo

0 10 20 30Weeks

40 50

Est

imat

ed m

ean

chan

ge fr

om b

asel

ine

for

BC

S

Figure 1. Estimated (a) FACT-B, (b) trial outcome index and (c) breast cancer subscale changes from baseline by treatment arm in a study of first-line therapywith paclitaxel-lapatinib (ErbB2þ subgroup)19.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

(Lþ P, n¼ 15; Pþ pla, n¼ 2). Limited conclusions can bedrawn due to the small sample size.

In the ErbB2þ subgroup, Q-TWiST differencesbetween groups ranged from 2 to 15 weeks, favoringLþ P over Pþ pla across all utility weight combinations(Table 3). For example, when the TOX and REL stateswere valued at half the utility of the TWiST state,the Q-TWiST difference was 8.5 weeks (p¼ 0.07).Figure 2(a) illustrates the combinations of utility weightsfor TOX and REL that result in statistically significantQ-TWiST differences between groups. Results from thesensitivity analysis using all AEs in the TOX state gavesimilar results (Figure 2(b)). We further examined theseresults using a slightly altered definition for ErbB2þ; usingthe IHC3þ results or FISHþ where IHC3þ status wasunknown provided 91 patients who were ErbB2þ (Lþ P,n¼ 52; Pþ pla, n¼ 39). Results were similar in this sub-group, with a Q-TWiST difference of more than 9 weeks(p¼ 0.05) when the TOX and REL states were valued athalf the utility of the TWiST state.

Discussion

In the ITT population QOL was stable during treatmentwith either lapatinib plus paclitaxel or paclitaxel mono-therapy. This finding is not unexpected since lapatinib isbelieved to have activity through ErbB2 inhibition. Thus,lack of any differences in clinical benefit in the ErbB2-negative and ErbB2-untested tumors would likely showno QOL benefits for the combination arm. The ErbB2þpopulation in this study made up 15% of the overall samplesize and QOL differences between groups were in the samedirection as in the ITT population. However, comparisonof FACT-B, TOI, and BCS scores in the ErbB2þ subgroupindicated significantly more favorable QOL results for thelapatinib plus paclitaxel patients than for patients takingpaclitaxel plus placebo. The addition of lapatinib to theregimen provided longer time to progression in ErbB2þMBC without a detrimental effect on QOL.

The Q-TWiST analysis was performed to further eval-uate treatment differences in patient QOL. Because themain goals for treating metastatic breast cancer are to pro-long life while improving or preserving quality of life ofpatients, a single measurement like Q-TWiST that com-bines both main objectives can be quite useful. Q-TWiSTallows the physician and patient to see at a glance whethertreatment toxicities outweigh the benefit. Unlike the ana-lysis of FACT-B scores, this analysis incorporated timeafter further progression, so it was not affected by missingQOL data after study treatment ended. Furthermore, the

1.0(a)

(b)1.0

0.8

0.8

0.6

0.6

0.4

0.4Utility for relapse

Util

ity fo

r to

xici

tyU

tility

for

toxi

city

0.2

0.20.0

0.0

1.0

1.0

0.8

0.8

0.6

0.6

0.4

0.4

Utility for relapse

0.2

0.20.0

0.0

Figure 2. Threshold utility plots (Q-TWiST differences in weeks) in theErbB2þ subgroup of a study of first-line therapy with lapatinib–paclitaxel.(a) TOX¼ grade 3/4 adverse events; (b). TOX¼ all AEsQ-TWiST: quality-adjusted time without symptoms of disease or toxicity of treatment.Positive numbers on slanted lines indicate more Q-TWiST for patients takinglapatinibþ paclitaxel versus paclitaxelþ placebo. Shading represents p-values fromthe test of Q-TWiST differences between groups: medium grey, p50.05; light, stripedband, 0.055p50.10; dark gray, p40.10.

Table 4. Estimated quality of life treatment differences* based on repeatedmeasures models with discontinuation scores assigned to next scheduledvisit in a study of first-line therapy with paclitaxel–lapatinib (ErbB2þsubgroup)19.

FACT-B scoreydifference(95% CI)

TOI scoreydifference(95% CI)

BCS scoreydifference(95% CI)

Week 9 3.4 (�4.2, 11.0) 2.5 (�2.8, 7.8) 2.0 (0.1, 4.0)Week 21 9.2 (0.6, 17.9) 6.4 (0.6, 12.1) 3.2 (1.3, 5.1)Week 33 6.7 (�2.1, 15.5) 5.0 (�1.0, 10.9) 2.1 (�0.3, 4.4)Week 45 11.8 (0.6, 23.0) 9.1 (1.4, 16.7) 2.6 (�0.8, 5.9)

*(lapatinibþ paclitaxel)� (paclitaxelþ placebo).yPositive difference in score indicates better QOL for lapatinibþ paclitaxelvs. paclitaxelþ placebo.BCS, breast cancer subscale; FACT-B, Functional Assessment of CancerTherapy-Breast; TOI, trial outcome index.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

Q-TWIST analysis directly assesses the impact of treat-ment toxicities on the patient experience. This is animportant consideration when combination therapy isintroduced, especially with an EGFR inhibitor sincethese agents are known to be associated with dermatologicevents. The increased rate of the serious events diarrhea,rash, and mucositis in the combination arm of the currentstudy have been reported previously23. The Q-TWiSTanalysis shows that these toxicity events are offset byextended time in the TWiST state. In the ErbB2þ sub-group, the combination treatment is expected to provide2–15 weeks of more quality-adjusted survival across utilityweightings. The advantage is most pronounced for patientswho rate the AE experience as better than time afterrelapse, but all patients are expected to benefit from thecombination treatment regardless of how they rate thevalue of periods with AEs.

Patient-reported utility data were not collected in thisstudy, so the threshold utility analyses provide a frameworkin which to assess expected Q-TWiST differences for arange of possible values. In a recent GSK study(EGF100151) in which EQ-5D scores were collectedfrom women with previously treated advanced or meta-static breast cancer, average utility value was 0.66 forTOX and 0.40 for the relapse period after further progres-sion24. At these valuations, lapatinib plus paclitaxel isexpected to provide more than 2 months more quality-adjusted survival than paclitaxel monotherapy forErbB2þ patients. These findings represent clinicallyimportant differences between treatment groups inquality-adjusted survival25.

It is quite common to have missing data for patient-reported questionnaires used in clinical trials. Patientsmay skip the entire assessment or only certain questionswithin the questionnaire. One limitation of this studydesign is that QOL was not assessed after disease progres-sion. This practice is typical in oncology clinical trials thatare designed to assess efficacy and safety, but it createsmissing QOL data that may be considered nonrandom.A pattern mixture model was conducted to assess theimpact of missing data on results. While this analysisgenerally provided confirmation of the findings from therepeated measures analysis, it suggested that treatmentdifference in QOL was most pronounced for the group ofpatients who progressed or withdrew from treatmentwithin 6 months. Sample sizes were too small to fullyexplore the impact of missing data configurations.

Lapatinib has previously been shown to increase time toprogression without negatively impacting QOL in pre-treated ErbB2þ MBC patients15. In the current study offirst-line MBC treatment, QOL scores were generally com-parable between groups in the overall (ITT) population;thus, supporting the idea that addition of lapatinib to thepaclitaxel regimen did not have a negative impact onQOL. A limitation of the FACT-B in this setting is that

the instrument may not be sensitive to gastrointestinaleffects of lapatinib, such as diarrhea. While acknowledgingthe discomfort that gastrointestinal problems may presentto the individual, the findings here suggest that these typ-ically low-grade effects did not have a large impact onaverage QOL. Where greatest efficacy was observed,among ErbB2þ patients, lapatinib plus paclitaxel demon-strated stable QOL compared with declining QOL inpatients taking paclitaxel alone. Some research is sugges-tive that the effect of tumor size reductions may be respon-sible for the better QOL outcomes among patients whorespond to lapatinib treatment15, but results may not begeneralizable across different patient populations at differ-ent disease stages. Additionally, analyses are hindered bylack of QOL assessments after progression occurs. In thisstudy, the Q-TWiST analysis provided evidence of anadvantage of combination therapy over monotherapy inthe ErbB2þ subgroup when considering the entire survivalperiod adjusted for relative utility of the health states.

Conclusion

QOL effects in the ErbB2þ subgroup were generally con-sistent with results in the ITT population, but with morepronounced differences between treatment groups, favor-ing the combination arm. The small number of ErbB2þpatients limits the generalizability of these results.Ongoing studies are assessing the combination of lapatinibwith paclitaxel versus paclitaxel alone (GSK Study#EGF104535) and the combination of lapatinib withpaclitaxel versus trastuzumab with paclitaxel (GSKStudy #EGF108919) as first-line treatments for ErbB2þMBC. Results from these studies could provide furtherconfirmation of the benefits of lapatinib in ErbB2þpatients and possibly extend this oral treatment optionto the first-line setting.

TransparencyThe EGF30001 study and analyses reported in this paper

were funded by GlaxoSmithKline. One employee fromGlaxoSmithKline participated with other co-authors in theinterpretation of analysis results and in manuscript writing/review.

Declaration of financial/other relationshipsM.A. is an employee of and holds stock in GlaxoSmithKline. B.S.and Y.W. are employees of RTI Health Solutions, a consultancycompany that also has research consultancy contracts withseveral other pharmaceutical companies. B.S., A.D.L., Y.W.and H.G. have disclosed as serving as consultants toGlaxoSmithKline. A.D.L. has disclosed receiving honorariafrom GlaxoSmithKline.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

AcknowledgmentsWe thank the patients who participated in the study and theirfamilies; the medical, nursing, and research staff at the studycenters; the independent data and safety monitoring committee;and the monitors, clinical operations staff, data managers, statis-ticians, and programmers at GlaxoSmithKline. We would alsolike to thank Michael Arbushites and Maria Koehler for theirvaluable assistance.

Previously presented in part at the 2008 San Antonio BreastCancer Symposium (SABCS) 10–14 December 2008, SanAntonio, Texas, and the 2009 American Society of ClinicalOncology (ASCO) annual meeting 29 May–2 June 2009,Orlando, FL, USA.

References1. Ferlay J, Bray F, Pisani P, et al. GLOBOCAN 2002: Cancer incidence, mortality

and prevalence. Worldwide IARC CancerBase No. 5. version 2.0. Lyon,

France: IARC Press, 2004

2. Ferlay J, Autier P, Boniol M, et al. Estimates of the cancer incidence and

mortality in Europe in 2006. Ann Oncol 2007;18:581-92

3. American Cancer Society. Cancer Facts & Figures 2008. Atlanta, GA: American

Cancer Society, 2008. Available at: http://www.cancer.org/downloads/

STT/2008CAFFfinalsecured.pdf [Last accessed 15 September 2008]

4. Meric F. HER-2/neu in the management of invasive breast cancer. J Am Coll

Surg 2002;194:488-501

5. Coleman RE. Skeletal complications of malignancy. Cancer 1997;80:

1588-94

6. Lin NU, Bellon JR, Winer EP. CNS metastases in breast cancer. J Clin Oncol

2004;22:3608-17

7. Lin NU, Winer EP. Brain metastases: the HER2 paradigm. Clin Cancer Res

2007;13:1648-55

8. Mundy GR. Metastasis to bone: causes, consequences and therapeutic

opportunities. Nat Rev Cancer 2002;2:584-93

9. Bernard-Marty C, Lebrun F, Awada A, et al. Monoclonal antibody-based tar-

geted therapy in breast cancer: current status and future directions. Drugs

2006;66:1577-91

10. Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapy plus a

monoclonal antibody against HER2 for metastatic breast cancer that over-

expresses HER2. N Engl J Med 2001;344:783-92

11. Marty M, Cognetti F, Maraninchi D, et al. Randomized phase II trial of the

efficacy and safety of trastuzumab combined with docetaxel in patients with

human epidermal growth factor receptor 2-positive metastatic breast cancer

administered as first-line treatment: the M77001 study group. J Clin Oncol

2005;23:4265-74

12. Montemurro F, Donadio M, Clavarezza M, et al. Outcome of patients with

HER2-positive advanced breast cancer progressing during trastuzumab-

based therapy. Oncologist 2006;11:318-24

13. Tripathy D, Slamon DJ, Cobleigh M, et al. Safety of treatment of metastatic

breast cancer with trastuzumab beyond disease progression. J Clin Oncol

2004;22:1063-70

14. Sherrill B, Amonkar M, Stein S, et al. Q-TWiST analysis of lapatinib combined

with capecitabine for the treatment of metastatic breast cancer. Brit J Cancer

2008;99:711-15

15. Zhou X, Cella D, Cameron D, et al. Lapatinib plus capecitabine versus

capecitabine alone for HER2þ (ErbB2þ) metastatic breast cancer:

quality-of-life assessment. Breast Cancer Res Treat 2009;117:577-89

16. O’Shaughnessy J, Blackwell KL, Burstein H, et al. A randomized study of

lapatinib alone or in combination with trastuzumab in heavily pretreated

ErbB2þ metastatic breast cancer progressing on trastuzumab therapy.

Abstract no. 1015. 2008 Annual Meeting of the American Society of

Clinical Oncology, Chicago, IL. 30 May-3 June 2008

17. Burstein HJ, Blackwell K, Wu Y, et al. Impact of lapatinib plus trastuzumab

versus single-agent lapatinib on quality of life (QOL) of patients with

trastuzumab-refractory ErbB2þ (ErbB2þ) metastatic breast cancer (MBC).

Abstract no. 154. 2008 Breast Cancer Symposium of the American Society of

Clinical Oncology, Washington, DC. 5-7 September 2008

18. Johnston S, Pippen Jr J, Pivot X, et al. Lapatinib combined with letrozole vs.

letrozole and placebo as first-line therapy for postmenopausal hormone

receptor positive metastatic breast cancer. J Clin Oncol 2009;27:5538-46

19. Di Leo A, Gomez HL, Aziz Z, et al. Phase III, double-blind, randomized

study comparing lapatinib plus paclitaxel with placebo plus paclitaxel

as first-line treatment for metastatic breast cancer. J Clin Oncol

2008;26:5544-52

20. Brady MJ, Cella DF, Mo F, et al. Reliability and validity of the Functional

Assessment of Cancer Therapy Breast quality-of-life instrument. J Clin

Oncol 1997;15:974-86

21. Eton DT, Cella D, Yost KJ, et al. A combination of distribution- and

anchor-based approaches determined minimally important differences

(MIDs) for four endpoints in a breast cancer scale. J Clin Epidemiol 2004;

57:898-910

22. Gelber RD, Goldhirsch A. A new endpoint for the assessment of adjuvant

therapy in postmenopausal women with operable breast cancer. J Clin Oncol

1986;4:1772-9

23. Ulhoa-Cintra A, Greenberg L, Geyer CE. The emerging role of lapatinib in

ErbB2-positive breast cancer. Curr Oncol Rep 2008;10:10-17

24. Sherrill B, Amonkar MM, Stein S, et al. Q-TWiST analysis of lapatinib com-

bined with capecitabine for the treatment of metastatic breast cancer. Br J

Cancer 2008;99:711-15

25. Revicki DA, Feeny D, Hunt TL, et al. Analyzing oncology clinical trial data using

the Q-TWiST method: clinical importance and sources for health state pref-

erence data. Qual Life Res 2006;15:411-23



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity L

ibra

ry U

trec

ht o

n 09

/29/

13Fo

r pe

rson

al u

se o

nly.

Documents

Quality-of-life and quality-adjusted survival (Q-TWiST) in patients receiving lapatinib in combination with paclitaxel as first-line treatment for metastatic breast cancer