European Journal of Cancer (2014) 50, 1925– 1941

A v a i l a b l e a t w w w . s c i e n c e d i r e c t . c o m

ScienceDirect

journa l homepag e : www.e j cancer . com

Review

Patient-reported outcomes in randomised controlledtrials of gynaecological cancers: Investigatingmethodological quality and impact on clinicaldecision-making

http://dx.doi.org/10.1016/j.ejca.2014.04.005

0959-8049/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author: Address: Head, Health Outcome Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), GIData Center, Via Benevento, 6, 00161 Rome, Italy. Tel.: +39 06 441 639831; fax: +39 06 4402516.

E-mail address: f.efficace@gimema.it (F. Efficace).

Fabio Efficace a,⇑, Marc Jacobs b, Andrea Pusic c, Elfriede Greimel d, Alfonso Piciocchi a,Jacobien M. Kieffer e, Alexandra Gilbert f, Peter Fayers g, Jane Blazeby h, On behalf ofEORTC Quality of Life Group (Patient Reported Outcome Measurements Over Time InONcology-PROMOTION Registry)

a Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italyb Academic Medical Center, University of Amsterdam, Department of Medical Psychology, Amsterdam, The Netherlandsc Department of Surgery, Memorial Sloan Kettering Cancer Center, New York, USAd Department of Obstetrics and Gynecology, Medical University of Graz, Graz, Austriae Division of Psychosocial Research & Epidemiology, The Netherlands Cancer Institute, Amsterdam, The Netherlandsf Leeds Institute of Cancer Studies and Pathology, University of Leeds, Leeds, UKg Institute of Applied Health Sciences, University of Aberdeen, Aberdeen, UKh Centre for Surgical Research, University of Bristol and University Hospitals NHS Foundation Trust, Bristol, UK

Received 5 February 2014; received in revised form 27 March 2014; accepted 4 April 2014Available online 10 May 2014

KEYWORDS

Gynaecologic cancerPatient-reported out-comesClinical trialsQuality of lifeClinical decision-making

Abstract Aim: The aim for this study is to investigate the methodological quality andpotential impact on clinical decision making of patient reported outcome (PRO) assessmentin randomised controlled trials (RCTs) in the gynaecological cancer sites.Methods: A systematic review identified RCTs published between January 2004 and June2012. Relevant studies were evaluated using a pre-determined extraction form which included:(1) Trial demographics and clinical and PRO characteristics; (2) level of PRO reporting and(3) bias, assessed using the Cochrane Risk of Bias tool. All studies were additionally analysedin relation to their relevance in supporting clinical decision making.Results: Fifty RCTs enrolling 24,991 patients were identified. In eight RCTs (16%) a PRO wasthe primary end-point. Twenty-one studies (42%) were carried out in a multi-national context.Where statistically significant PRO differences between treatments were found, it related in

MEMA

1926 F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941

most cases to both symptoms and domains other than symptoms (n = 17, 57%). The majorityof studies (n = 42, 84%) did not mention the mode of administration nor the methods of col-lecting PRO data. Statistical approaches for dealing with missing data were only explicitlymentioned in nine RCTs (18%). Sixteen RCTs (32%) were considered to be of high-qualityand thus able to inform clinical decision making. Higher-quality PRO studies were generallyassociated with RCTs that were at a low risk of bias.Conclusion: This study showed that RCTs with PROs were generally well designed and con-ducted. In a third the information was very informative to fully understand the pros and consof PROs treatment decision-making.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Gynaecological cancers arise from the cervix, ovary,endometrium, vulva or vagina and as well as affecting2.2% of the female population by the age of 65 theyare the second commonest cause of cancer death inwomen after breast cancer [1].

Gynaecological cancers are treated with differenttreatment modalities including surgery, radiotherapyand chemotherapy either alone or in combination.Whilst combinations of different treatments are knownto improve survival they also increase the risk of side-effects in both the short and long term with patientscontinuing to report more gastrointestinal and sexualdysfunction symptoms than women in the generalpopulation in the years following treatment [2]. Thesesymptoms are associated with considerable impairmentin health-related quality of life (HRQOL) [3–5] thatcan also persist over the long-term period [6].

It is increasingly recognised that a comprehensiveevaluation of treatment effectiveness should include apatient reported outcome (PRO) assessment to fullycapture patients’ perceptions of symptoms, functioningand general well-being [7]. Inclusion of PROs can beparticularly valuable in randomised controlled clinicaltrials (RCTs) as it can potentially generate unique datato help health physicians to make more informed treat-ment decisions. However, information derived fromPROs need to be based on well-planned RCTs to ensurethat results are solid enough to robustly inform clinicalpractice [8]. Several methodological aspects should befully considered when implementing PROs in a RCTsetting [9]. Some excellent examples of the importantinformation that can be drawn by including PROs ingynaecological RCTs are available. To illustrate, thePost Operative Radiation Therapy in Endometrial Car-cinoma (PORTEC) trial found a significant reduction inthe rate of local regional recurrence with the addition ofpost operative radiotherapy to standard surgical treat-ment of endometrial cancer. However, this reductiondid not translate into an overall survival advantageand patients treated with the additional radiotherapyreported higher rates of gastrointestinal toxicity [10].

The use of HRQOL in this RCT led to the decision torecommend postoperative radiotherapy only in high-risk patients where the risk of relapse is considered tooutweigh the potential treatment related toxicity [10].Although this represents a concrete example of PROimplementation in a RCT setting, previous data haveshown a number of methodological flaws, which havehampered drawing strong conclusions from many RCTsconducted in several cancer disease sites [11–13].

While previous work has investigated the methodo-logical quality of studies in cervical cancer survivors, itreported studies published up to 2005 and was notfocused on RCTs [6]. On this ground, the broad scopeof this systematic review was to provide evidence-baseddata to help health care policy makers and physiciansbetter appreciate the state-of-art of PRO research inmain gynaecological cancers. Specific objectives were:(1) assess the quality of PRO reporting and methodolog-ical quality of each RCT, (2) identify high-quality PROstudies most likely to inform clinical decision-making (3)synthesise main clinical and PRO findings from the highquality studies.

2. Materials and methods

2.1. Search strategy for identification of studies

A systematic literature search was conducted inMedline, the Cochrane Controlled Trials Register,PsycINFO, and PsycARTICLES to identify RCTspublished between January 2004 and June 2012. Thesearch strategy was restricted to RCTs. Only Englishlanguage articles were considered and no restrictionswere included in the search field description. Expertsin the field were contacted to identify potentiallyrelevant articles not retrieved in the electronic search.Key searching strategy is reported in the onlineAppendix A. Titles and abstracts of identified articleswere screened for inclusion. Additional publicationswere identified by scanning reference lists of relevantarticles. Details on searching strategy and selectionprocess were documented according to the PRISMAguidelines [14].

F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941 1927

2.2. Selection criteria

2.2.1. Types of participants

Clinical trials in adult patients with a diagnosis ofcervical, ovarian, endometrial, vaginal or vulval cancerwere included regardless of disease stage. Studies ofpatients undergoing screening or with benign diseasewere excluded.

2.2.2. Types of intervention

All RCTs comparing conventional treatments wereincluded. Studies considering psychosocial interventionsor complementary therapies were not eligible.

2.2.3. Types of outcome measures examined

Studies including a PRO as a primary or secondaryoutcome were included. We included both RCTs thatevaluated either multi-dimensional patient-reportedHRQOL or other types of health outcomes as long asthey were reported by patients themselves. Studiesevaluating treatment satisfaction and adherence totherapy were not included.

2.2.4. Types of studies

We included all RCTs that (1) compared differentconventional treatment modalities and symptomsmanagement, and (2) had enrolled at least 50 patients.We restricted our review to RCTs because they arethe gold standard of research by which decisions aremade regarding the clinical effectiveness of treatments.Studies including a heterogeneous sample of cancersites were included if dealing with gynaecologicalcancers. Conference abstracts and case reports wereexcluded.

2.3. Data extraction and type of information extracted

Data extracted from the included RCTs were storedin predefined electronic data extraction forms (eDEFs)[15] and a web-based data collection system wasdeveloped for the purpose of this research (http://pro-motion.gimema.it/). Three reviewers (M.J., J.K. andA.P.) independently extracted data on: (1) basic trialdemographics (e.g. year, journal), (2) clinical and PROcharacteristics (e.g. number of patients enrolled, studylocation, treatments being compared, PRO instrumentused and clinical and PRO assessments). Summary offindings (i.e. PROs and clinical) was also extracted; (3)the quality of PRO reporting, based on the recently pub-lished guidelines by the International Society for Qualityof Life Research (ISOQOL) [16] and 4) risk of bias usingthe Cochrane Risk of Bias tool [17]. Results were cross-checked, and discrepancies were resolved by discussionwith the senior author (F.E.). In trials with multiplepublications, we obtained relevant data by combiningall trial publications.

2.4. Quality assessment and identification of high quality

studies

PRO quality assessment was based on the recentlydeveloped ISOQOL consensus-based recommendationsfor reporting of PROs in RCT publications [16]. Theseguidelines currently represent the highest quality criteriaavailable and they form the basis for the recentlypublished PRO extension of the CONSORT (CONsoli-dated Standards of Reporting Trials) guidelines [8]. TheISOQOL guidelines comprise up to 29 key criteria(depending on whether the PRO is a secondary orprimary end-point of the trial) that a study should doc-ument in order for the PRO data to be reliable. Each cri-terion was rated as either ‘yes’ (scored as 1) if the issuewas addressed or ‘no’ (scored as 0) if not. The higherthe score the higher the quality. In order to identifyhigh-quality studies that may have an impact on clinicaldecision-making, we a priori determined that at leasttwo thirds of the ISOQOL recommended criteria mustbe satisfied. Additional details on methodology usedhave been previously reported [15]. To assess internalvalidity (i.e. freedom from bias) of RCTs, we alsoapplied criteria from the Cochrane Risk of Bias tool.We thus evaluated each RCT for its adequacy ofsequence generation; allocation concealment; blindingof participants/personnel; completeness of outcomedata; blinding of outcome assessment; no selectivity ofoutcome reporting and other sources of bias [17].

3. Results

The systematic literature search yielded 2735 records(Fig. 1). Application of inclusion criteria identified 50RCTs enrolling overall 24,991 patients publishedbetween January 2004 and June 2012 (see Appendixfor the full list of papers retrieved). The majority ofthese (i.e. 32%) have been conducted on patients withovarian cancer.

3.1. Study demographics and PRO assessment

In eight RCTs (16%) a PRO was the primary end-point. Twenty-one studies (42%) were carried out in amulti-national context. Twenty-six studies (52%) wereat least partially supported by industry. Twenty-threestudies (46%) involved patients with mixed diseasestages, and 16 (32%) only recruited patients withadvanced/metastatic disease. Although 31 RCTs (62%)enrolled more than 200 patients, PROs were the primaryend-point in only two of these RCTs. The two most fre-quently used PRO instruments, used alone or in con-junction with other measures, were the EuropeanOrganization for Research and Treatment of Cancer(EORTC) QLQ-C30 in 24 RCTs (48%), and the

Fig. 1. Schematic breakdown of literature search results of Gynaecological Randomised Controlled Trials (preferred reporting items for systematicreviews and metaanalysis). PRO = patient-reported outcome.

1928 F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941

Functional Assessment of Cancer Therapy (FACT)instruments in 16 (32%). Thirty studies (60%) found aPRO difference between treatment arms. When a PROdifference was found, this related to both symptomsand domains other than symptoms in 17 RCTs (57%).Further details are provided in Table 1.

3.2. Overview on PRO assessment methodology

The majority of publications (42 RCTs; 84%) did notreport the mode of administration of the PRO tool norwere the methods of collecting the data described. Inaddition, 43 RCTs (86%) did not state a PRO hypothesisnor did they specify to which PRO domain the hypoth-esis was relevant. Although the extent of missing datawas well documented in 30 RCTs (60%), only nineRCTs (18%) explicitly stated the statistical approachesapplied for dealing with missing data. In addition, 15RCTs (30%) explained the reasons for missing data. Six-teen RCTs (32%) provided a flow diagram that provideda description of the allocation of participants and thoselost to follow-up for PROs specifically. When discussingtheir findings, only 15 RCTs (30%) discussed the clinicalsignificance of their findings. Moreover, only 14 RCTs(28%) discussed the generalisability issues that wereuniquely related to the PRO results. The limitations ofthe PRO components of the trial were discussed in lessthan half of the RCTs included (46%). A complete over-view is provided in Table 2.

3.3. Overview on outcomes from higher-quality PRO

studies

Sixteen RCTs (32%) that were likely to providerobust PRO data to inform clinical decision-making(Table 3) were identified. Five RCTs (31%) focused onadvanced/metastatic disease, three (19%) reported onnon-metastatic disease stages and seven (44%) includedon mixed disease stages (both loco-regional and meta-static disease). One RCT did not explicitly state patients’disease stage. The majority of these studies were at a lowrisk of bias for sequence generation (n = 11, 69%), allo-cation concealment (n = 11, 69%), blinding of partici-pants/personnel (n = 3, 19%) and outcome (n = 11,69%), attrition (n = 10, 63%) and selective outcomereporting (n = 10, 62%). Conversely, the percentages oflow risk of bias RCTs were generally a bit lower in lowerquality PRO studies. Fig. 2 depicts risk of bias of allRCTs, classified by the PRO quality rating (high versuslow).

3.3.1. Summary outcomes from metastatic/advanced

disease

The largest RCT, with 976 patients [18–23], had pro-gression free survival as the primary end-point and usedthe EORTC QLQ-C30 and the ovarian specific module(OV-28) to assess HRQOL. This trial showed that theexperimental arm had significant improvements in pro-gression free survival and less peripheral neuropathy,

Table 1Randomised controlled trial (RCT) demographic characteristics.

Variable Patient-reportedoutcome (PRO)end-pointn (%)

Totaln (%)

Primary8 (16%)

Secondary42 (84%)

Basic RCT demographics

Anatomic site of cancer Cervical 2 (25) 7 (16.675) 9 (18)Endometrial 2 (25) 6 (14.29) 8 (16)Ovarian 1 (12.50) 15 (35.72) 16 (32)Multiple disease sites 3 (37.50) 14 (33.33) 17 (34)

International No 6 (75) 23 (54.76) 29 (58)Yes 2 (25) 19 (45.24) 21 (42)

Industry supported (fully or in part)* No 7 (87.5) 17 (40.48) 24 (48)Yes 1 (12.5) 25 (59.52) 26 (52)

Overall study sample size (regardless of patientsincluded in the PRO analysis)

6200 patients 6 (75) 13 (30.95) 19 (38)>200 patients 2 (25) 29 (69.05) 31 (62)

Disease stage Advanced/metastatic 2 (25) 14 (33.33) 16 (32)Locoregional/no distant metastasis 5 (62.5) 5 (11.9) 10 (20)Mixed disease stages 1 (12.5) 22 (52.38) 23 (46)Unclear 0 (0) 1 (2.38) 1 (2)

Broad treatment type Radiotherapy 2 (25) 7 (16.67) 9 (18)Surgery 4 (50) 6 (14.29) 10 (20)Chemotherapy 3 (37.50) 36 (85.71) 39 (78)Target therapy 2 (25) 1 (2.38) 3 (6)

Difference between treatment arms in the primaryend-point

No 0 (0) 20 (47.62) 20 (40)Yes 8 (100) 22 (52.38) 30 (60)

Overall survival (OS) difference favouringexperimental treatment

No 1 (12.5) 28 (66.67) 29 (58)Yes 0 (0) 7 (16.67) 7 (14)N/A (in case OS was not assessed) 7 (87.5) 7 (16.67) 14 (28)

PRO-related basic characteristics

PRO instrument used European Organization for Research andTreatment of Cancer (EORTC) instruments

3 (37.50) 21 (50) 24 (48)

Functional Assessment of Cancer Therapy(FACT) instruments

2 (25) 14 (33.33) 16 (32)

Visual analogue scale (VAS) 2 (25) 1 (2.38) 3 (6)Others 1 (12.5) 6 (14.29) 7 (14)

PRO difference between treatment arms No differences at all 1 (12.5) 18 (42.86) 19 (38)Yes broadly favouring experimental treatment� 7 (87.5) 12 (28.57) 19 (38)Yes broadly favouring standard treatment� 0 (0) 11 (26.19) 11 (22)N/A 0 (0) 1 (2.38) 1 (2)

If statistically significant PRO difference exists, inwhich domain?

Symptoms only 3 (42.86) 6 (26.09) 9 (30)PRO domains other than symptoms only (e.g.functional scales or global QoL)

0 (0) 4 (17.39) 4 (13.33)

Both domains (symptoms + domains other thansymptoms)

4 (57.14) 13 (56.52) 17(56.67)

Length of PRO assessment during RCT Up to 6 months 8 (100) 15 (35.71) 23 (46)Up to 1 year 0 (0) 13 (30.95) 13 (26)More than 1 year 0 (0) 14 (33.33) 14 (28)

Secondary paper on PRO� No 7 (87.5) 30 (71.43) 37 (74)Yes 1 (12.5) 12 (28.57) 13 (26)

* Assessed if explicitly stated or if one or more authors were affiliated to a pharmaceutical company. This evaluation is based solely on informationextracted from the paper.� Often, multiple PRO domains (e.g. from multidimensional HRQOL questionnaires) are analysed at the same time in longitudinal PRO–RCTs; toillustrate, difference in such domains might favour the experimental treatment arm at a given time-point and then favouring the control treatmentarm at a different time point over the course of the study. So, the term ‘broadly’ was inserted to account for this possible discrepancy.� Assessed as ‘yes’ if at least one paper was published in addition to the original RCT report.

F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941 1929

Table 2Level of patient-reported outcomes (PROs) reporting by type of end-point (PRO primary versus secondary end-point of the trial).

PRO end-pointn (%)

Totaln (%)

Primary8 (16%)

Secondary42 (84%)

Title and abstractThe PRO should be identified as an outcome in the abstract No 2 (25) 10 (23.81) 12 (24)

Yes 6 (75) 32 (76.19) 38 (76)The title of the paper should be explicit as to the randomised controlled trial (RCT) including a PRO No 4 (50) – 4 (50)

Yes 4 (50) – 4 (50)

Introduction, background and objectivesThe PRO hypothesis should be stated and should specify the relevant PRO domain if applicable No 7 (87.5) 36 (85.71) 43 (86)

Yes 1 (12.5) 4 (9.52) 5 (10)N/A (if explorative) 0 (0) 2 (4.76) 2 (4)

The introduction should contain a summary of PRO research that is relevant to the RCT No 3 (37.5) – 3 (37.5)Yes 5 (62.5) – 5 (62.5)

Additional details regarding the hypothesis should be provided, including the rationale for the selected domain(s),the expected direction(s) of change and the time points for assessment

No 8 (100) – 8 (100)

MethodsOutcomes

The mode of administration of the PRO tool and the methods of collecting data should be described No 7 (87.5) 35 (83.33) 42 (84)Yes 1 (12.5) 7 (16.67) 8 (16)

Electronic mode of PRO administration* No 1 (12.5) 7 (16.67) 8 (16)N/A 7 (87.5) 35 (83.33) 42 (84)

The rationale for choice of the PRO instrument used should be provided No 8 (100) 34 (80.95) 42 (84)Yes 0 (0) 8 (19.05) 8 (16)

Evidence of PRO instrument validity and reliability should be provided or cited No� 4 (50) 8 (19.05) 12 (24)Yes 4 (50) 34 (80.95) 38 (76)

The intended PRO data collection schedule should be provided No 2 (25) 6 (14.29) 8 (16)Yes 6 (75) 36 (85.71) 42 (84)

PROs should be identified in the trial protocol post hoc analyses should be identified No 6 (75) 28 (66.67) 34 (68)Yes 2 (25) 14 (33.33) 16 (32)

The status of PRO as either a primary or secondary outcome should be stated No 1 (12.5) 0 (0) 1 (2)Yes 5 (62.5) 39 (92.86) 44 (88)Unclear 2 (25) 3 (7.14) 5 (10)

A citation for the original development of the PRO instrument should be provided No� 5 (62.5) – 5 (62.5)Yes 3 (37.5) – 3 (37.5)

Windows for valid PRO responses should be specified and justified as being appropriate for the clinical context No 7 (87.5) – 7 (87.5)Yes 1 (12.5) – 1 (12.5)

Sample size

There should be a power sample size calculation relevant to the PRO based on a clinical rationale No 5 (62.5) – 5 (62.5)Yes 3 (37.5) – 3 (37.5)

Statistical methods

1930F

.E

ffica

ceet

al./E

uro

pea

nJ

ou

rna

lo

fC

an

cer5

0(

20

14

)1

92

5–

19

41

There should be evidence of appropriate statistical analysis and tests of statistical significance for each PRO hypothesis tested Yes 1 (12.5) 4 (9.52) 5 (10)N/A (If PRO hypotheses werenot stated)

7 (87.5) 38 (90.48) 45 (90)

The extent of missing data should be stated� No 3 (37.5) 17 (40.48) 20 (40)Yes 5 (62.5) 25 (59.52) 30 (60)

Statistical approaches for dealing with missing data should be explicitly stated � No 7 (87.5) 34 (80.95) 41 (82)Yes 1 (12.5) 8 (19.05) 9 (18)

The manner in which multiple comparisons have been addressed should be provided No 5 (62.5) – 5 (62.5)Yes 3 (37.5) – 3 (37.5)

ResultsParticipant flow

A flow diagram or a description of the allocation of participants and those lost to follow-up should be provided for PROsspecifically

No 6 (75) 28 (66.67) 34 (68)Yes 2 (25) 14 (33.33) 16 (32)

The reasons for missing data should be explained No 3 (37.5) 32 (76.19) 35 (70)Yes 5 (62.5) 10 (23.81) 15 (30)

Baseline data

The study patients characteristics should be described including baseline PRO scores No 3 (37.5) 16 (38.1) 19 (38)Yes 5 (62.5) 26 (61.9) 31 (62)

Outcomes and estimation

Are PRO outcomes also reported in a graphical format?* No 4 (50) 22 (52.38) 26 (52)Yes 4 (50) 20 (47.62) 24 (48)

The analysis of PRO data should account for survival differences between treatment groups if relevant No 1 (12.5) – 1 (12.5)N/A (if not relevant) 7 (87.5) – 7 (87.5)

Results should be reported for all PRO domains(if multi-dimensional)and items identified by the reference instrument No 2 (25) – 2 (25)Yes 6 (75) – 6 (75)

The proportion of patients achieving pre-defined responder definitions should be provided where relevant No 4 (50) – 4 (50)Yes 1 (12.5) – 1 (12.5)N/A (if not relevant) 3 (37.5) – 3 (37.5)

DiscussionLimitations

The limitations of the PRO components of the trial should be explicitly discussed No 6 (75) 21 (50) 27 (54)Yes 2 (25) 21 (50) 23 (46)

Generalisability

Generalisability issues uniquely related to the PRO results should be discussed No 2 (25) 34 (80.95) 36 (72)Yes 6 (75) 8 (19.05) 14 (28)

Interpretation

Are PRO interpreted? (Not only re-stated)* No 3 (37.5) 19 (45.24) 22 (44)Yes 5 (62.5) 23 (54.76) 28 (56)

The clinical significance of the PRO findings should be discussed No 7 (87.5) 28 (66.67) 35 (70)Yes 1 (12.5) 14 (33.33) 15 (30)

Methodology used to assess clinical significance (in case this was addressed)* Anchor based 0 (0) 3 (21.43) 3 (20)Distribution based 1 (100) 9 (64.29) 10

(66.67)Other 0 (0) 2 (14.29) 2

(13.33)(continued on next page)

F.

Effi

cace

eta

l./Eu

rop

ean

Jo

urn

al

of

Ca

ncer

50

(2

01

4)

19

25

–1

94

11931

Tab

le3

(co

nti

nu

ed)

PR

Oen

d-p

oin

tn

(%)

To

tal

n(%

)

Pri

mar

y8

(16%

)S

eco

nd

ary

42(8

4%)

Th

eP

RO

resu

lts

sho

uld

be

dis

cuss

edin

the

con

text

of

the

oth

ercl

inic

altr

ial

ou

tco

mes

No

2(2

5)20

(47.

62)

22(4

4)Y

es6

(75)

22(5

2.38

)28

(56)

Oth

erin

form

atio

nP

roto

col

Aco

py

of

the

inst

rum

ent

sho

uld

be

incl

ud

edif

ith

asn

ot

bee

np

ub

lish

edp

revi

ou

sly

(It

cou

ldb

efo

un

din

the

arti

cle

Ap

pen

dix

or

inth

eo

nli

ne

vers

ion

)N

o8

(100

)–

8(1

00)

Fo

rd

escr

ipti

vep

urp

ose

s,su

bh

ead

ings

of

this

tab

lere

flec

tth

ato

fre

po

rted

inth

eIn

tern

atio

nal

So

ciet

yfo

rQ

ual

ity

of

Lif

eR

esea

rch

(IS

OQ

OL

)re

com

men

ded

stan

dar

ds,

ho

wev

er,

rati

ng

of

item

sw

asin

dep

end

ent

of

loca

tio

no

fth

ein

form

atio

nw

ith

inth

em

anu

scri

pt;

N/A

:n

ot

app

lica

ble

.–,

Ind

icat

esit

ems

that

are

no

tap

pli

cab

leas

thes

ear

ere

com

men

ded

tob

ere

po

rted

on

lyw

hen

PR

Ois

ap

rim

ary

end

-po

int.

*T

he

foll

ow

ing

item

sh

ave

no

tb

een

incl

ud

edin

the

ISO

QO

Lre

com

men

ded

stan

dar

ds

[16]

bu

th

ave

bee

nad

ded

inth

ista

ble

toh

ave

aw

ider

ou

tlo

ok

on

the

leve

lo

fre

po

rtin

g.�

We

eval

uat

edas

‘no

’if

all

PR

Om

easu

res

use

din

the

stu

dy

wer

en

ot

vali

dat

ed.

�T

hes

eit

ems

wer

eo

rigi

nal

lyco

mb

ined

inth

eIS

OQ

OL

reco

mm

end

edst

and

ard

s[1

6]b

ut

hav

eb

een

spli

tin

this

rep

ort

tob

ette

rin

vest

igat

ep

oss

ible

dis

crep

anci

esb

etw

een

do

cum

enta

tio

no

fP

RO

mis

sin

gd

ata

(i.e

.re

po

rtin

gh

ow

man

yp

atie

nts

did

no

tco

mp

lete

agi

ven

qu

esti

on

nai

reat

any

give

nti

me

po

int)

vers

us

actu

alre

po

rtin

go

fst

atis

tica

lm

eth

od

sto

add

ress

this

issu

e.A

lso

,w

ew

ante

dto

be

con

sist

ent

wit

hit

ems

rep

ort

edin

the

CO

Nso

lid

ated

Sta

nd

ard

so

fR

epo

rtin

gT

rial

s(C

ON

SO

RT

)P

RO

exte

nsi

on

(i.e

.st

ati

stic

al

ap

pro

ach

esfo

rd

eali

ng

wit

hm

issi

ng

da

taar

ere

po

rted

asa

stan

dal

on

eis

sue)

.

1932 F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941

and other chemotherapy side-effects, and less impact onbody image than standard chemotherapy, althoughthese benefits were not translated into a difference inglobal quality of life scores. Other smaller RCTs thatused the FACT instruments did not show a cleardifference in HRQOL between chemotherapy-basedtreatment arms also.

3.3.2. Summary outcomes from non-metastatic disease

stageThe largest trial, with 429 patients, conducted by

Armstrong and colleagues [24–27] showed that intrave-nous therapy, when compared to intraperitonealtherapy, improved progression free survival. In addi-tion, after intravenous therapy, patients showed morephysical and functional well-being, less ovarian cancersymptoms and less abdominal discomfort using theFACT questionnaires.

3.3.3. Summary outcomes from mixed disease stage

The biggest study included in this systematic review,with 2616 patients overall, was conducted by Walkerand colleagues [28–30]. This trial randomised patientswith clinical stage I to stage IIA uterine cancer tolaparoscopy or open laparotomy with recurrence freesurvival as the primary outcome. This trial showed thatpatients treated by laparoscopy had fewer moderate tosevere post-operative adverse events, and a lower fre-quency of hospitalisation of more than 2 days. In addi-tion, results obtained mainly with the use of the FACTquestionnaires, showed that these patients had betterphysical functioning, body image, less pain and an ear-lier resumption of normal activities and return to workat 6 weeks after surgery. Further details and a summaryof key findings from other studies are reported inTable 3.

4. Discussion

In this systematic review, whose data has been gath-ered by the PROMOTION Registry, a large interna-tional initiative by the EORTC Quality of Life Group,all PRO-based RCTs conducted in gynaecological can-cers have been searched. The broad goal of our reviewwas to summarise the main findings stemming from thisstudy, to help health care policy makers and physiciansbetter appreciate the state-of-art in this field.

Some 25,000 patients with various gynaecologicalcancers have been enrolled in 50 RCTs over the lastfew years and, of these, at least one-third has providedreliable PRO data and has considered the informationalongside the clinical data to make treatment recom-mendations. We thus recommend that future trials con-tinue to do this and that PRO trial design and reportingis conducted in a robust manner so that patients can beinformed how treatments impact not only on survival

Table 3Randomised controlled trials with robust PRO design: basic study characteristics.

Study* International Age of patients (years)� Overall

study

sample

size

Baseline

PRO

sample

size

PRO instruments used Primary end-

point

Treatment outline Summary main clinical results Summary of PRO results/PRO

treatment recommendations

Anatomic

site of

cancer

Metastatic/advanced disease stage

Kurtz JE et al. Ann

Oncol

2011;22(11):2417–

23;

Alexandre J et al.

Br J Cancer

2012;106(4):633–7;

Brundage M et al.

Ann Oncol 2012;

23(8):2020–7;

Gladieff L et al.

Ann Oncol 2012;

23(5):1185–9;

Joly et al.; Gynecol

Oncol

2011;122(2):226–

32;

Pujade-Lauraine

et al. J Clin Oncol

2010; 28(20):3323–

9

Yes Median (range)

Carboplatin–pegylated

liposomal doxorubicin

(CD): 60.5 (24–82)

Carboplatin–paclitaxel

(CP): 61 (27–82)

976 879 EORTC QLQ-C30;

EORTC QLQ-OV28

Progression

free-survival

CD arm: combination of pegylated

liposomal doxorubicin (PLD) (30 mg/

m2 intravenously on day 1) and

carboplatin (AUC 5 based on the

Calvert formula using glomerular

filtration rate calculated from serum

creatinine values according to the

method of Cockroft and Gault,

administered intravenously on day 1 at

4-week intervals.

CP arm: combination of paclitaxel

(175 mg/m2 intravenously on day 1) and

carboplatin (AUC 5 intravenously on

day 1) at 3-week intervals. Random

assignment was performed in permuted

blocks of six cycles

With medi follow-up of 22 months,

Progressio ree-Survival for the

pegylated osomal doxorubicin with

carboplati rm was statistically

superior to he carboplatin and

paclitaxel (hazard ratio, 0.821; 95%

confidence terval (CI) = 0.72–0.94;

p = 0 .005 edian progression free

survival (P ) was 11.3 versus

9.4 months espectively. Overall severe

non-haema logic toxicity (36.8%

versus 28.4 ; p < 0.01) leading to early

discontinu on (15% versus 6%;

p < 0.001) curred more frequently in

the CP arm

Global quality of life (QoL) and

abdominal symptom scores

improved over time in both

arms; at 6 months, 36% of

patients met criteria for

improved symptoms. Treatment

with CD resulted in less

peripheral neuropathy (9.8

versus 24.2), fewer other

chemotherapy side-effects (9.5

versus 16.2) and less impact on

body image (3.8 versus 10.4)

versus CP (all p < 0.02) at

6 months

Multiple

disease sites

Cella D et al. Gynecol

Oncol

2010;119(3):531–7;

Monk BJ et al. J

Clin Oncol

2009;27(28): 4649–

55

No Median (range)

Vinorelbine (VC) = 49;

(24–76) gemcitabine

(GC) = 45; (20–89)

Topotecan (TC) = 48;

(25–75) paclitaxel

(PC) = 50; (29–81)

513 410 FACT-Cx TOI; FACT/

GOG-NTX; BPI

Overall

survival

Patients were randomly assigned to

paclitaxel 135 mg/m2 over 24 h plus Cis

50 mg/m2 day 2 every 3 weeks (PC,

reference arm); vinorelbine 30 mg/

m2 days 1 and 8 plus Cis 50 mg/m2 day

1 every 3 weeks (VC); gemcitabine

1000 mg/m2 day 1 and 8 plus Cis 50 mg/

m2 day 1 every 3 weeks (GC); or

topotecan 0.75 mg/m2 days 1, 2 and 3

plus Cis 50 mg/m2 day 1 every 3 weeks

(TC).

Duration: up to 6 months

No statisti lly significant difference No statistically significant

differences in health-related

quality of life (HRQOL),

neuropathy, or pain

Cervical

Long HJ et al. J Clin

Oncol 2005;23(21):

4626–33;

Monk BJ et al. J

Clin Oncol

2005;23(21): 4617–

25;

Long HJ et al.

Gynecol Oncol

2006;100(3): 537–

43

No Cisplatin–topotecan

(CT): 46(22–84)

Cisplatin (CPT): 48 (27–

76)

364 284 FACT-G, FACT-Cx,

FACT/GOG-NTX, BPI,

UNI

Overall

survival

Cisplatin 50 mg/m2 every 3 weeks

(CPT);

Cisplatin 50 mg/m2 IV day 1 plus

topotecan 0.75 mg/m2 days 1–3 every

21 days (CT) (methotrexate 30 mg/

m2 days 1, 15 and 22, vinblastine 3 mg/

m2 days 2, 15 and 22, doxorubicin

30 mg/m2 day 2, and cisplatin 70 mg/

m2 day 2 every 28 days.

Methotrexate, vinblastine, doxorubicin

and cisplatin (MVAC). Max six cycles

Patients re iving cisplatin and

topotecan had statistically superior

outcomes those receiving cisplatin

CPT, with edian overall survival of 9.4

and 6.5 mo ths (p = 0 .017), median

PFS of 4.6 nd 2.9 months (p = 0.014)

and respon rates of 27% and 13%,

respectivel

There was no statistical evidence

suggesting that reported QoL

and adverse effects scores

changed over time differently

across regimens.

Baseline FACT-G (p = 0.0016)

and BPI (p = 0.0001) scores were

significantly associated with

patient age; older patients had

better QOL and less pain.

Baseline UNI was positively

correlated with FACT-G

(r = 0.66; p < 0.001) and Cx

subscale (r = 0.29; p < 0.001),

and negatively related to BPI

(r = �0.41; p < 0.0001). Baseline

FACT-Cx (FACT-G + Cx

subscale) was associated with

survival

Cervical

(continued on next page)

F.

Effi

cace

eta

l./Eu

rop

ean

Jo

urn

al

of

Ca

ncer

50

(2

01

4)

19

25

–1

94

11933

of

an

n F

lip

n a

t

arm

in

); m

FS

, r

to

%

ati

oc

ca

ce

CT

to

m

n

a

se

y

Table 3 (continued))

Study* International Age of patients (years)� Overall

study

sample

size

Baseline

PRO

sample

size

PRO instruments used Primary end-

point

Treatment outline Summary of m in clinical results Summary of PRO results/PRO

treatment recommendations

Anatomic

site of

cancer

Moore DH et al. J

Clin Oncol 2004;

22(15):3113–9;

McQuellon RP

et al. Gynecol

Oncol 2006;

101(2):296–304

No Median (range)

Cisplatin (C): 46.0 (22–

84)

Cisplatin plus paclitaxel

(CP): 48.5 (21–77)

280 252 FACT-G; FACT-Cx; A

Trial Outcome Index

representing Physical Well

Being (PWB) + Functional

Well-Being (FWB) + the

Cervix subscale;

BPI-SF; neurotoxicity

subscale

Overall

survival;

progression

free-survival;

Objective

Response

Rate

Cisplatin (C): IV dose of 50 mg/m2 at

the rate of 1 mg/min.

Cisplatin plus paclitaxel (CP): paclitaxel

IV dose of 135 mg/m2 as a 24-h infusion

followed immediately by cisplatin at a

dose of 50 mg/m2 Duration every

3 weeks for six cycles

Objective resp ses occurred in 19%

(6% complete us 13% partial) of

patients receiv g cisplatin versus 36%

(15% complet lus 21% partial)

receiving cispl in plus paclitaxel

(p = 0.002). T median (PFS) was 2.8

and 4.8 month respectively, for

cisplatin versu cisplatin plus paclitaxel

(p < 0.001)

There was no significant

difference in QOL scores.

The BPI-SF revealed a decline in

pain scores in both arms from

the first to fourth assessments.

The rate of QOL drop-out for

any reason was higher for C

(53%) compared to CP (38%)

(p < 0.05). At the fourth time

point, 60% of living patients in

both arms completed a QOL

assessment

Cervical

Fleming GF et al. J

Clin Oncol 2004;

22(11):2159–66

No 650, N = 22

51–60, N = 72

61–70, N = 110

71–80, N = 54

P81, N = 5

273 216 FACT/GOG-NTX Overall

survival

Doxorubicin cisplatin (AP) doxorubicin

60 mg/m2 and cisplatin 50 mg/m2.

Doxorubicin cisplatin paclitaxel (TAP)

doxorubicin 45 mg/m2 and cisplatin

50 mg/m2 (day 1), followed by paclitaxel

160 mg/m2 (day 2) with filgrastim

support every 3 weeks to a maximum of

seven cycles

Objective resp se (57% versus 34%;

p < 0.01), PFS median, 8.3 versus

5.3 months; p 0 .01) and OS (median,

15.3 versus 12 months; P = 0.037)

were improve ith TAP

Following two cycles of

chemotherapy, patients on the

TAP arm reported a significantly

higher neurotoxicity score than

did patients on the AP arm.

Significant differences in the

mean score were observed, and

sustained following the second

cycle of chemotherapy.

Furthermore, patients reported

neurotoxicity increased

significantly during the

treatment period in the TAP

arm, but not in the AP arm

Endometrial

Non-metastatic disease stage

Armstrong DK et al.

N Engl J Med

2006; 354(1):34–

43;

Wenzel LB et al. J

Clin Oncol 2007;

25(4):437–43;

von Gruenigen VE

et al. Gynecol

Oncol 2012;124(3):

379–82;

Krivak TC et al.

Gynecol Oncol

2009; 115(1):81–5

No Intravenous-Therapy

(IV) (N = 210) and

Intraperitoneal-Therapy

(IP) (N = 205)

21–30 0; 4(2)

31–40 15 (7); 8 (4)

41–50 43 (20);

52 (25)

51–60 74 (35); 62 (30)

61–70 56 (27); 53 (26)

71–80 19 (9); 24 (12)

>80 3 (1); 2 (1)

429 399 FACT-O; FACT-TOI;

FACT-G subscales (PWB,

Physical Well Being; FWB,

Functional Well-Being,

SWB, Social well-being and

EWB, Emotional Well-

being);

FACT/GOG NTX

Progression

free-survival

and overall

survival

Intravenous-therapy (IV): 135 mg of

intravenous paclitaxel per square metre

of body-surface area over a 24-h period

on day 1 followed by 75 mg of

intravenous cisplatin per square metre

on day 2.

Intraperitoneal-therapy (IP): 135 mg of

intravenous paclitaxel per square metre

over a 24-h period on day 1 followed by

100 mg of intraperitoneal cisplatin per

square metre on day 2 and 60 mg of

intraperitoneal paclitaxel per square

metre on day 8 duration: every 3 weeks

for six cycles

The median d ation of progression-

free survival i he intravenous-therapy

and intraperit eal-therapy groups was

18.3 and 23.8 onths, respectively

(p = 0.05).

The median d ation of overall survival

in the intraven us-therapy and

intraperitonea erapy groups was 49.7

and 65.6 mon , respectively (p = 0.03)

Quality of life was significantly

worse in the intraperitoneal-

therapy group before cycle 4 and

three to six weeks after

treatment but not one year after

treatment.

Physical and functional well-

being and ovarian cancer

symptoms were significantly

worse in the IP arm before cycle

4 (p < 0.001) and 3–6 weeks after

treatment (p = 0.001 for FACT-

TOI).

Patients in the IP arm also

reported significantly worse

abdominal discomfort (AD)

before cycle 4 (p = 0.001) and

significantly worse Ntx 3–

6 weeks (p = 0.001) and

12 months (p = 0.003) after

completing IP treatment. In

general, however, the quality of

life of both groups improved

over time

Multiple

disease sites

1934F

.E

ffica

ceet

al./E

uro

pea

nJ

ou

rna

lo

fC

an

cer5

0(

20

14

)1

92

5–

19

41

a

on

pl

in

e p

at

he

s,

s

on

(

<

.3

d w

ur

n t

on

m

ur

o

lth

ths

Janda M et al.

Contemp Clin

Trials

2006;27(4):353–63;

Janda M et al.

Lancet Oncol 2010;

11(8):772–80

Yes Total laparoscopic

hysterectomy (TLH):

mean (SD) 62.8 (10.0)

Total abdominal

hysterectomy (TAH):

Mean (SD) 62.7 (9.7)

361 332 FACT-G; EnWB (FACT

subscale Endometrial)

(FACT-

G + EnWB = FACT-

Endometrial) Body image

scale; EQ-5D (EuroQoL-

VAS)

Patient

Reported

Outcomes

(PRO)

(including

QOL or

symptoms

relief)

QoL at

6 months

Total laparoscopic hysterectomy

(TLH).

Total abdominal hysterectomy (TAH)

Operating me was significantly longer

in the TL group (138 min [SD 43])

than in th TAH group (109 min [SD

34]; p = 0 1).

Postopera ely, twice as many patients

in the TA group experienced adverse

events of ade 3 or higher (33 of 142

[23.2%] v us 22 of 190 [11.6%] in the

TLH gro p = 0.004).

Postopera e serious adverse events

occurred re in the TAH group (27 of

142 [19.0% than in the TLH group (16

of 190 [7. ]; p = 0.002)

Patients who had TLH reported

significantly greater

improvement in QoL from

baseline compared with those

that had TAH, in all subscales

apart from emotional and social

well-being. Improvements in

QoL up to 6 months after

surgery continued to favour

TLH, except in the emotional

and social well-being measures

of the FACT and the visual

analogue scaled of the EuroQol-

VAS.

The greatest differences were

noted in FWB (13% greater

improvement for patients with

TLH), PWB (11%), EnWB (6%),

and the overallFACT-G

summary score (7%; p = 0.001

for all comparisons).

Patients in the TLH group also

reported a 5% (p = 0.001)

greater improvement in body

image and 7.5% (p = 0.001)

greater improvement in overall

QoL (EuroQoL-VAS) than

patients in the TAH group.

During the late post-op recovery

phase (3–6 mths after surg),TLH

patients recovered significantly

more in their physical,

(p = 0.008), functional

(p = 0.009), endometrial cancer-

specific (p = 0.003) and overall

wellbeing (FACT-G; p = 0.03),

and also had better QoL

recovery with regard to body

image (p = 0.001)

Endometrial

Mourits MJ et al.

Lancet Oncol 2010;

11(8):763–71;

Bijen CB et al.

Gynecol Oncol

2011; 121(1):76–82

No Total abdominal

hysterectomy (TLH):

n = 185 median: 62 range

40–89;

Total abdominal

hysterectomy (TAH):

n = 94 median 63 range

39–86

283 Unclear SF-36, SAQ, BIS, VAS for

general health perception.

EQ-5D

Major

complication

rate

Total abdominal hysterectomy (TAH)

and bilateral salpingo-oophorectomy

versus Total laparoscopic hysterectomy

(TLH) and bilateral salpingo-

oophorectomy

The prop ion of major complications

was 14�6% 7 of 185) in the TLH group

versus 14 (14 of 94) in the TAH

group, wi a diff erence of –0.3% (95%

CI = �9. o 8.5; p = 0.95).TLH was

associated ith signifi cantly less blood

loss (p < 001), less use of pain

medicatio p < 0.0001), a shorter

hospital s (p < 0.0001), and a faster

recovery 0.002), but the procedure

took long than TAH (p < 0.0001)

A higher percentage of patients

in TLH group (76.3%) resumed

daily activities after 6 weeks

than in TAH group (62.2%).

Patients who had TLH scored

significantly higher on the

physical functioning subscale of

the SF-36 at 6 weeks, and on the

role-physical subscale at

3 months after the procedure.

Patients who had TAH scored

significantly higher on the

vitality subscale of the mental

dimension 3 months after

surgery

Endometrial

(continued on next page)

F.

Effi

cace

eta

l./Eu

rop

ean

Jo

urn

al

of

Ca

ncer

50

(2

01

4)

19

25

–1

94

11935

ti

H

e

.00

tiv

H

gr

ers

up;

tiv

mo

])

9%

ort

(2

�9%

th

1 t

w

0.0

n (

tay

(p =

er

Table 3 (continued))

Study* International Age of patients (years)� Overall

study

sample

size

Baseline

PRO

sample

size

PRO instruments used Primary end-

point

Treatment outline Summary of main clinical results Summary of PRO results/PRO

treatment recommendations

Anatomic

site of

cancer

Mixed disease stage (loco-regional disease and metastatic disease)

Walker JL et al. J Clin

Oncol 2009;

27(32):5331–6;

Kornblith AB

et al. J Clin Oncol

2009; 27(32):5337–

42;

Walker JL et al. J

Clin Oncol 2012;

30(7):695–700

Yes Median (Range)

Laparoscopy

62.8 (55.4–71.6)

Laparotomy

62.7 (54.9–70.6)

2616 727 FACT-G; AP

MOS-SF36; PF

BPI; BI

Recurrence

free survival

Patients with clinical stage I to IIA

uterine cancer were randomly assigned

to laparoscopy (n 1696) or open

laparotomy (n 920), including

hysterectomy, salpingo-oophorectomy,

pelvic cytology and pelvic and para-

aortic lymphadenectomy

Laparoscopy had fewer moderate to

severe postoperative adverse events than

laparotomy (14% versus 21%,

respectively; p < 0.0001); significantly

longer operative time (median, 204

versus 130 min, respectively; p < 0.001).

Hospitalisation of more than 2 days was

significantly lower in laparoscopy versus

laparotomy patients (52% versus 94%,

respectively; p < 0.0001).

Pelvic and para-aortic nodes were not

removed in 8% of laparoscopy patients

and 4% of laparotomy patients

(p < 0.0001)

In an intent-to-treat analysis,

laparoscopy patients reported

significantly higher Functional

Assessment of Cancer Therapy –

General (FACT-G) scores

(p = 0.001), better physical

functioning (p = 0.006), better

body image (BI; p < 0.001), less

pain (p < 0.001) and its

interference with QoL

(p < 0.001), and an earlier

resumption of normal activities

(p = 0.003) and return to work

(p = 0.04) over the 6-week

postsurgery period, as compared

with laparotomy patients.

By 6 months, the only difference

between arms was the better BI

in laparoscopy

Endometrial

Greimel ER et al. J

Clin Oncol 2006;

24(4):579–86;

du Bois AJ et al. J

Natl Cancer Inst.

2003;95(17):1320–

9.

Hilpert F et al. Ann

Oncol 2007;

18(2):282–7

Yes Mean 56.7, Standard

deviation (SD) 10.93;

range 20.8–77.4 and 57.7

SD 10.11; range 25.4–

83.6 in the paclitaxel plus

carboplatin (TC) and

paclitaxel plus cisplatin

(PT)

798 TC: 397

eligible,

366

forms

returned

PT: 386

eligible,

357

forms

returned

EORTC QLQ-C30 Progression

free-survival

Paclitaxel (185 mg/m2) + carboplatin

(AUC = 6) (TC arm) versus paclitaxel

(185 mg/m2) + cisplatin 75 mg/m2 (PT

arm) – all six courses per 3 weeks

The proportion of patients without

progression at 2 years was not

statistically significantly different

between the two treatment arms (40.0%

for PT versus 37.5% for TC,

difference = 2.5%, one-sided 95%

confidence interval [CI] = �1 to 8.2%).

Median progression-free survival time

in the TC arm (17.2 months, 95%

CI = 15.2–19.3 months) and the PT arm

(19.1 months, 95% CI = 16.7–

21.5 months) were also not statistically

significantly different; the same was true

of median overall survival time

(43.3 months, 95% CI = 37.2–

47.8 months versus 44.1 months, 95%

CI = 40.2–49.4 months, for the TC and

PT arms, respectively)

Patients in the TC arm showed

better means scores after

treatment on overall QoL

(p = 0.012), physical functioning

(p = 0.012), role functioning

(P = 0.005) and cognitive

functioning (p = 0.024),

compared with the PT arm.

Concerning symptom

experience, patients undergoing

TC showed less nausea and

vomiting (p < 0.001), less

appetite loss (p < 0.001) and less

fatigue (p = 0.033) after

completion of treatment

compared with patients

undergoing PT

Ovarian

Carey MS et al.

Gynecol Oncol

2008; 108(1):100–

5;

Butler L et al. J

Clin Oncol 2004;

22(12):2461–8;

Piccart MJ J Natl

Cancer Inst 2000;

92(9):699–708;

Bezjak A et al. J

Clin Oncol

2004;22(22):4595–

603

Yes Median (Range)

Cyclophosphamide (CP):

58 (22–85)

paclitaxel (TP): 58 (23–

79)

680 152 EORTC QLQ-C30 Progression

free-survival

TP arm = paclitaxel at a dose of

175 mg/m2 as a 3-h infusion followed by

cisplatin at a dose of 75 mg/m2

CP arm = cyclophosphamide at

750 mg/m2 followed by cisplatin at

75 mg/m2. Duration every 3 weeks

between 3 and 9 cycles

At a median follow-up of 38.5 months

and despite a high rate of crossover

(48%) from the cyclophosphamide arm

to the paclitaxel arm at first detection of

progression of disease, a longer

progression-free survival (log-rank

p = 0.0005; median of 15.5 months

versus 11.5 months) and a longer overall

survival (log-rank p = 0.0016; median of

35.6 months versus 25.8 months) were

seen in the paclitaxel regimen compared

with the cyclophosphamide regimen

Clinically meaningful

improvements compared with

baseline (change scores > 10)

were found in both arms during

the treatment period in a

number of domains and items,

including global QOL,

emotional function, social

function, fatigue, pain, sleep,

constipation, appetite,

abdominal swelling and

abdominal cramps.

Improvements in global QOL

persisted for the duration of

follow-up. More neurosensory

effects and myalgia were found

in the paclitaxel arm

Ovarian

1936F

.E

ffica

ceet

al./E

uro

pea

nJ

ou

rna

lo

fC

an

cer5

0(

20

14

)1

92

5–

19

41

Wenzel L et al. J Clin

Oncol 2005;

23(24):5605–12;

Rose PG et al. N

Engl J Med

2004;351(24):2489–

97.

No Median (range)

Secondary surgery plus

chemotherapy group:

58.1 (25.4–81.6)

Chemotherapy-alone

group: 57.0 (27.0–81.6)

550 376 FACT-G; FACT-O; GOG

(supplemental questions)

Overall

survival

Name drugs: paclitaxel + cisplatin

Doses: Chemo only: paclitaxel 135 mg/

m2 over 24 h followed by immediately

by cisplatin Cis 75 mg/m2;

Duration: maximum of six cycles of

3 weeks

Surgery + chemo: paclitaxel 135 mg/m2

over 24 h followed immediately by

cisplatin Cis 75 mg/m2 every 3 weeks for

max six cycles. Secondary cytoreduction

performed as soon as possible after

haematologic recovery of the 3rd cycle,

but within six weeks after the

completion of the third cycle

No statistically significant differences For all patients, QOL decreased

approximately 1 unit from the

first to second assessment.

Significant improvement

observed at 6 months (p < 0.001)

was sustained at 12 months, with

no appreciable between-group

difference (p = 0.048). The

baseline FACT-O score was

associated with overall survival

(p = 0.048) but not progression-

free survival. Less neurotoxicity

was reported among patients

who did (38.4%) versus did not

(54.0%) undergo interval

secondary cytoreduction at the

third assessment (p = 0.005) and

older patients experienced more

long-term effects

Ovarian

Nout RA et al. Lancet

2010;

375(9717):816–23;

Nout RA et al. Eur

J Cancer

2012;48(11):1638–

48;

Nout RA et al. J

Clin Oncol 2009;

27(21):3547–56

No Pelvic external beam

radiotherapy

(EBRT) = Median 69

(SD:7);

vaginal brachytherapy

(VBT) = Median:70

(SD:7)

427 Unclear EORTC QLQ-C30

Version 3.0; subscales for

bowel and bladder

symptoms from the prostate

cancer module (PR25) and

subscale for sexual

functioning and symptoms

from Ovarian Cancer

Module (OV28)

Vaginal

recurrence

Vaginal brachytherapy (VBT) = 1.

High–dose-rate (HDR). schedules

aimed at 45–50 Gy by schedules of

21 Gy in three fractions of 7 Gy, 1 week

apart for the high-dose rate;

2. Low-dose rate (LDR) schedules:

30 Gy at 50–70 cGy/h

3. Medium-dose-rate. (MDR)

schedules: 28 Gy at 100 cGy/h in one

session. pelvic external beam

radiotherapy (EBRT) = A dose of

46 Gy 2 Gy fractions, five times per

week, duration unclear

No significant differences between arms

regarding vaginal recurrence.

Rates of acute grade 1–2

gastrointestinal toxicity were signifi

cantly lower in the vaginal

brachytherapy VBT group than in the

pelvic external beam radiotherapy

EBRT group at completion of

radiotherapy (12.6% [27/215] versus

53.8% [112/208]

Patients in the VBT group

reported better social

functioning (p < 0.002) and

lower symptom scores for

diarrhoea, faecal leakage, the

need to stay close to the toilet

and limitation in daily activities

because of bowel symptoms

(p < 0.001). At baseline, 15% of

patients were sexually active;

this increased significantly to

39% during the first year

(p < 0.001). Sexual functioning

and symptoms did not differ

between the treatment groups

Endometrial

Randall ME et al. J

Clin Oncol 2006;

24(1):36–44;

Bruner DW et al.

Qual Life Res

2007;16(1):89–100

No Median = 63 422 317 FS; APN; FACE; FACT-G. Progression

free-survival

Whole-abdominal irradiation (WAI):

irradiation dose was 30Gyin 20 daily

fractions. After WAI, patients received

a boost to the true pelvis or to an

extended field encompassing pelvic

lymph nodes (PLNs) and positive para-

aortic lymph nodes (PALNs). The boost

dose was 15 Gy in eight fractions. All

fields were treated once daily, 5 days per

week.

AP: chemotherapy; doxorubicin 60 mg/

m2 plus cisplatin 50 mg/m2 every

3 weeks for seven cycles, followed by

one cycle of cisplatin.

The stage-adjusted hazard ratio for

progression was 0.71 favouring

doxorubicin-cisplatin AP (95%

CI = 0.55–0.91; p < 0.01). At

60 months, 50% of patients receiving AP

were predicted to be alive and disease

free adjusting for stage compared with

38% of patients receiving whole-

abdominal irradiation WAI. The stage-

adjusted death hazard ratio was 0.68

(95% CI = 0.52–0.89; p < 0.01)

favouring AP. At 60 months and

adjusting for stage, 55% of AP patients

were predicted to be alive compared

with 42% of WAI patients

WAI patients reported worse FS

(p < 0.001) and FACE

(p < 0.001) scores at end of

treatment and poorer FACE

scores 3 months post-treatment

(p = 0.004) compared to AP

patients. APN scores were

significantly worse among AP

patients at end of treatment, and

3 and 6 months post-treatment

(p < 0.001 for all). FACT-G

scores did not differ between the

two arms at any assessment

point.

Endometrial

(continued on next page)

F.

Effi

cace

eta

l./Eu

rop

ean

Jo

urn

al

of

Ca

ncer

50

(2

01

4)

19

25

–1

94

11937

Table 3 (continued))

Study* International Age of patients (years)� Overall

study

sample

size

Baseline

PRO

sample

size

PRO instruments used Primary end-

point

Treatment outline Summary of m in clinical results Summary of PRO results/PRO

treatment recommendations

Anatomic

site of

cancer

Wilkinson PM et al.

Br J Cancer 2006;

94(7):947–54

Yes Mean: Epoetin

alfa = 59.1 (±10.6)

(range 35–87)

Best standard

treatment = 60.3 (±11.2)

(range (30–79)

182 102 FACT-G, FACT-An,

CLAS; LASA

Changes in

haemoglobin

(Hb) level

from baseline

to study end

Epoetin alfa 10,000–20,000 IU three

times weekly plus best standard

treatment (BST = transfusion of red

blood cells, as needed)/BST only. The

planned duration of study treatment

was a maximum of 28 weeks, which

included 18– 24 weeks of chemotherapy

(maximum, six cycles) plus up to

4 weeks after the last chemotherapy

dose

For the epoet alfa group, mean Hb

increased by 1 g dl(raise to the -

1power) by w ks 4–6 and was

significantly in eased from baseline

through study nd (p < 0.001). The

mean change Hb from baseline was

significantly ( 0.001) greater for

epoetin alfa th n BST patients at all

postbaseline e luations. Significantly

fewer epoetin fa than BST patients

required trans sion(s) after the first

4 weeks of tre ment (7.9 versus 30.5%;

p < 0.001)

Significant differences from

baseline favouring epoetin alfa

over BST for all three CLAS

change scores (Energy Level,

Ability to Do Daily Activities,

Overall QOL) and the average

median CLAS change score

during chemotherapy

Ovarian

Unclear disease stage

Alvarez Secord A et al.

Cancer

2012;118(13):3283–

93;

Pokrzywinski R

et al. Gynecol

Oncol

2011;123(3):505–

10;

Havrilesky LJ

et al. Cancer 2012;

118(2):386–91

No Mean/median (SD)

(range)

Docetaxel and

Carboplatin (cDC): 63.8/

64 (10.2) (43–84)

Sequential

Docetaxel and

Carboplatin (sDC): 63.0/

64.5 (10.0) (39–82)

150 148 FACT-O Progression

free-survival

Name drugs: docetaxel and carboplatin

Doses: cDC: docetaxel (30 mg/m2 iv on

days 1 and 8) combined with

carboplatin (AUC = 6 mg/mL/minute

iv on day 1) every 3 weeks.

sDC: docetaxel (30 mg/m2 iv on days 1

and 8) every 3 weeks followed by

carboplatin at an AUC of six

intravenously every 3 weeks at first

progression or after six cycles of

docetaxel for stable disease or a partial

response.

Duration: six cycles of 3 weeks

The median p gression-free survival

(PFS) was 13. onths (95% confidence

interval [CI] = .9–16.8) for

combination d cetaxel and carboplatin

cDC and 8.4 nths (95% CI = 7.1–

11.0) for sequ tial docetaxel and

carboplatin sD

Sequential docetaxel followed by

carboplatin (sDC) was

associated with significant

improvements in the FACT-O

TOI (p = 0.013), FACT-O total

score (p = 0.033) and ovarian

cancer-specific (OCS) module

(p = 0.029) compared to the

combination docetaxel and

carboplatin group (cDC)

Multiple

disease sites

Abbreviations: EORTC QLQ-C30: European Organization for Research and Treatment of Cancer Quality of Life Questionnaire Core-30; EO TC QLQ-OV28: European Organization for Researchand Treatment of Cancer Quality of Life Questionnaire-Ovarian Module-28; AUC: area under the curve; FACT-G: Functional Assessmen of Cancer Therapy – General; FACT-Cx: FunctionalAssessment of Cancer Therapy – Cervix; TOI: Trial Outcome Index; FACT/GOG-NTX: Functional Assessment of Cancer Therapy/Gynaeco gic Oncology Group – Neurotoxicity four-item scale;UNI = UNISCALE; BPI-SF: Brief Pain Inventory – short form; FACT-O: Functional Assessment of Cancer Therapy – Ovarian; EQ-5D European Quality of Life-5 Dimensions; VAS: visualanalogue scale; SF-36: The Medical Outcomes Study Short Form 36-Item Health Survey; SAQ: the Sexual Activity Questionnaire; BIS: body image scale; MOS-SF36 PF: the Physical FunctioningSubscale of the Medical Outcome Study – short form; AP: additional treatment related symptoms; BI: body image; APN: assessment for eripheral neuropathy scale; FS: fatigue scale; FACE:functional alterations due to changes in elimination; FACT-An: the fact-anaemia; CLAS: cancer linear analogue scale; LASA: linear analogu scale assessment; PRO: patient-reported outcome.* No ranking was made based on level or reporting in this table. Studies are ordered by overall number of patients recruited in the study.� Reported data as available in the paper.

1938F

.E

ffica

ceet

al./E

uro

pea

nJ

ou

rna

lo

fC

an

cer5

0(

20

14

)1

92

5–

19

41

a

in

.8

ee

cr

e

in

p <

a

va

al

fu

at

ro

7 m

9

o

mo

en

C

Rtlo

:

pe

Fig. 2. Risk bar chart showing risk of bias across randomised controlled trials (RCTs) by quality of patient-reported outcome (PRO) studies.

F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941 1939

and toxicity data but also on aspects that are essential totheir overall health and well being.

Including a PRO assessment in a RCT requires spe-cial consideration of a number of methodologicalaspects, and major efforts have been made to increasestandards of PRO reporting in RCT by internationalleading groups such as the CONSORT Group and theISOQOL [8,16]. With regard to the key items recentlysuggested by the CONSORT PRO extension [8], ourwork has highlighted specific areas most in need ofurgent improvement, such as the importance of docu-menting statistical methods used to handle missing data.While 60% of studies indeed documented the extent ofPRO missing data, only 18% further reported detailson how these were managed in the analysis. This is inline with what has been found in PRO-based RCTs inprostate cancer [15].

Another important result, previously identified insimilar analyses [15], is the association between higherquality PRO studies and lower risk of RCT bias(Fig. 2). This might support the concept that large andwell-designed trials, hence probably those that havebeen designed and led by expert methodological andclinical groups who have also received adequate finan-cial and practical support, are more likely to incorporatewell planned PRO assessment. However, this should beconfirmed in other cancer disease sites.

Another important issue that arose from this reviewis the importance of selecting an appropriate PRO mea-sure when designing an RCT. Several of the robustPRO-based RCTs used a site specific PRO instrumentalongside a generic one and nearly all used the EORTCand FACT instruments. Where cancer generic measuresmay not be sensitive enough, site-specific instruments

maybe better able to detect clinically meaningfulchanges in HRQOL, and allow for a better discrimina-tion between treatment arms. This is critical for treat-ment of a disease that has such an impact on personalaspects of health (e.g. sexual function and body image)as well as the traditional measures. It turns out to beespecially important with gynaecologic cancer therapiesbecoming increasingly tailored to individual risk factors,and with the evolution of new biological agents present-ing a prospect of maintenance therapy, possibly extend-ing treatment side-effects. These new therapeutic regimesmay have the risk of additional toxicities, and willhighlight the importance of selecting PRO instrumentstailored to the research questions. For gynaecologiconcology specific instruments have been developed forthe major cancer sites (ovarian, endometrial, cervical,vulva) that can be used in clinical trials. With the avail-ability of valid and sensitive site-specific measures,HRQOL measurement will continue to gain prominenceas a principal outcome measure in clinical trials.

The choice of PRO instruments is critical in RCTs asthis should be as sensitive as possible to detect changesin the specific research question being tested [31]. Thus,an important question that needs to be further investi-gated in future studies is the relationship betweenPRO results and questionnaires used. Whether PROresults are linked to the quality and suitability of theinstrument for the specific question or are more corre-lated to the quality of the trial in general, remains tobe clarified. We aim to further examine this issue withinour EORTC QLG PROMOTION Registry.

This review has limitations. Although we used a com-prehensive searching strategy, it is still possible thatsome RCTs with a PRO component might have been

1940 F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941

missed. Also, as previously noted [15], our definition of‘high quality studies’ is somewhat arbitrary and does notconsider the relevance for PRO inclusion in the specificRCT context. Lastly, it should be noted that possiblehigh quality HRQOL reports published after the cut-off date of this systematic literature search, are notconsidered in current work. To illustrate, Greimel andcolleagues recently published a comprehensive HRQOLanalysis [32] of an RCT [33] that we reviewed. There-fore, the additional HRQOL information stemmingfrom this RCT could not be included. This is a moregeneral issue of RCTs with a HRQOL component,because, the large amount of data collected typicallyprevents the possibility of including all information intoa single paper. Hence, a separate publication onHRQOL analysis and outcomes is necessary to allowfor a critical appraisal of the robustness of HRQOLfindings in medical practice. Ideally, this additional pub-lication should be in the same journal to the one report-ing the main RCT findings in order to reach the samereaders and possibly have a major clinical impact.

This paper also has several notable strengths. PROmethodological evaluation was based on the most solidand up to date quality criteria [16]. Furthermore, everyRCT has been reviewed at least by two independentreviewers, permitting a calibrated assessment of allPRO-based RCTs in gynaecological cancer research.Also, this review not only provided data on methodo-logical aspects of PRO assessment but also synthesisedthe clinical and PRO outcomes stemming for higherquality studies in an attempt to provide medical commu-nity with concrete take home messages.

To conclude, quite a few RCTs have been conductedin patients with gynaecological cancers over the last fewyears and in at least some one-third of these, PRO out-comes have been very informative to fully understandpro and cons of the new treatment approaches beingtested. Investigators should pay particular attention tothe most frequently unmet methodological aspects iden-tified in this work to further improve the quality andtransparency of their PRO findings in future RCTpublications.

Conflict of interest statement

None declared.

Acknowledgements

This paper stems from a larger project (i.e. PatientReported Outcome Measurements Over Time InONcology-PROMOTION Project) funded by a researchgrant from the European Organisation for Research andTreatment of Cancer (EORTC) Quality of Life Group.Also, additional support for the conduct of the studywas provided by the Italian Group for Adult Hemato-

logic Diseases (GIMEMA). JMB is supported by theMRC ConDuCT Hub for Trials MethodologyResearch. We also acknowledge Alessandro Perrecaand Salvatore Soldati, from the GIMEMA, for theircontribution to data management.

Appendix A. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at http://dx.doi.org/10.1016/j.ejca.2014.04.005.

References

[1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM.Estimates of worldwide burden of cancer in 2008: GLOBOCAN2008. Int J Cancer 2010;127(12):2893–917. http://dx.doi.org/10.1002/ijc.25516.

[2] Bergmark K, Avall-Lundqvist E, Dickman PW, Henningsohn L,Steineck G. Vaginal changes and sexuality in women with ahistory of cervical cancer. N Engl J Med 1999;340(18):1383–9.http://dx.doi.org/10.1056/NEJM199905063401802.

[3] Greimel ER, Winter R, Kapp KS, Haas J. Quality of life andsexual functioning after cervical cancer treatment: a long-termfollow-up study. Psychooncology 2009;18(5):476–82. http://dx.doi.org/10.1002/pon.1426.

[4] Janda M, Obermair A, Cella D, Crandon AJ, Trimmel M. Vulvarcancer patients’ quality of life: a qualitative assessment. Int JGynecol Cancer 2004;14(5):875–81. http://dx.doi.org/10.1111/j.1048-891X.2004.14524.x.

[5] Barker CL, Routledge JA, Farnell DJ, Swindell R, Davidson SE.The impact of radiotherapy late effects on quality of life ingynaecological cancer patients. Br J Cancer2009;100(10):1558–65. http://dx.doi.org/10.1038/sj.bjc.6605050.

[6] Vistad I, Fossa SD, Dahl AA. A critical review of patient-ratedquality of life studies of long-term survivors of cervical cancer.Gynecol Oncol 2006;102(3):563–72. http://dx.doi.org/10.1016/j.ygyno.2006.03.050, pii: S0090-8258(06)00316-7.

[7] Basch E, Abernethy AP, Mullins CD, et al. Recommendations forincorporating patient-reported outcomes into clinical compara-tive effectiveness research in adult oncology. J Clin Oncol2012;30(34):4249–55. http://dx.doi.org/10.1200/JCO.2012.42.5967.

[8] Calvert M, Blazeby J, Altman DG, Revicki DA, Moher D,Brundage MD. Reporting of patient-reported outcomes inrandomized trials: the CONSORT PRO extension. JAMA2013;309(8):814–22. http://dx.doi.org/10.1001/jama.2013.879.

[9] Chassany O, Sagnier P, Marquis P, et al. Patient-reportedoutcomes: the example of health related quality of life-a Europeanguidance document for the improved integration of health relatedquality of life assessment in the drug regulatory process. Drug InfJ 2002;36:209–38.

[10] Creutzberg CL, van Putten WL, Koper PC, et al. Surgery andpostoperative radiotherapy versus surgery alone for patients withstage-1 endometrial carcinoma: multicentre randomised trial.PORTEC Study Group. Post operative radiation therapy inendometrial carcinoma. Lancet 2000;355(9213):1404–11, pii:S0140673600021395.

[11] Efficace F, Bottomley A, van Andel G. Health related quality oflife in prostate carcinoma patients: a systematic review ofrandomized controlled trials. Cancer 2003;97(2):377–88. http://dx.doi.org/10.1002/cncr.11065.

[12] Claassens L, van Meerbeeck J, Coens C, et al. Health-relatedquality of life in non-small-cell lung cancer: an update of asystematic review on methodologic issues in randomized con-

F. Efficace et al. / European Journal of Cancer 50 (2014) 1925–1941 1941

trolled trials. J Clin Oncol 2011;29(15):2104–20. http://dx.doi.org/10.1200/JCO.2010.32.3683.

[13] Bottomley A, Therasse P. Quality of life in patients undergoingsystemic therapy for advanced breast cancer. Lancet Oncol2002;3(10):620–8, pii: S1470204502008768.

[14] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reportingitems for systematic reviews and meta-analyses: the PRISMAstatement. PLoS Med 2009;6(7):e1000097. http://dx.doi.org/10.1371/journal.pmed.1000097.

[15] Efficace F, Feuerstein M, Fayers P, et al. Patient-reportedoutcomes in randomised controlled trials of prostate cancer:methodological quality and impact on clinical decision making.Eur Urol 2013. http://dx.doi.org/10.1016/j.eururo.2013.10.017,pii: S0302-2838(13)01090-7.

[16] Brundage M, Blazeby J, Revicki D, et al. Patient-reportedoutcomes in randomized clinical trials: development of ISOQOLreporting standards. Qual Life Res 2013;22(6):1161–75. http://dx.doi.org/10.1007/s11136-012-0252-1.

[17] Higgins JP, Altman DG, Gotzsche PC, et al. The CochraneCollaboration’s tool for assessing risk of bias in randomised trials.BMJ 2011;343:d5928. http://dx.doi.org/10.1136/bmj.d5928.

[18] Kurtz JE, Kaminsky MC, Floquet A, et al. Ovarian cancer inelderly patients: carboplatin and pegylated liposomal doxorubicinversus carboplatin and paclitaxel in late relapse: a GynecologicCancer Intergroup (GCIG) CALYPSO sub-study. Ann Oncol2011;22(11):2417–23. http://dx.doi.org/10.1093/annonc/mdr001.

[19] Alexandre J, Brown C, Coeffic D, et al. CA-125 can be part of thetumour evaluation criteria in ovarian cancer trials: experience ofthe GCIG CALYPSO trial. Br J Cancer 2012;106(4):633–7.http://dx.doi.org/10.1038/bjc.2011.593.

[20] Brundage M, Gropp M, Mefti F, et al. Health-related quality oflife in recurrent platinum-sensitive ovarian cancer – results fromthe CALYPSO trial. Ann Oncol 2012;23(8):2020–7. http://dx.doi.org/10.1093/annonc/mdr583.

[21] Gladieff L, Ferrero A, De Rauglaudre G, et al. Carboplatin andpegylated liposomal doxorubicin versus carboplatin and paclit-axel in partially platinum-sensitive ovarian cancer patients: resultsfrom a subset analysis of the CALYPSO phase III trial. AnnOncol 2012;23(5):1185–9. http://dx.doi.org/10.1093/annonc/mdr441.

[22] Joly F, Ray-Coquard I, Fabbro M, et al. Decreased hypersensi-tivity reactions with carboplatin–pegylated liposomal doxorubicincompared to carboplatin–paclitaxel combination: analysis fromthe GCIG CALYPSO relapsing ovarian cancer trial. GynecolOncol 2011;122(2):226–32. http://dx.doi.org/10.1016/j.ygyno.2011.04.019, pii: S0090-8258(11)00316-7.

[23] Pujade-Lauraine E, Wagner U, Aavall-Lundqvist E, et al. Pegy-lated liposomal Doxorubicin and Carboplatin compared withPaclitaxel and Carboplatin for patients with platinum-sensitiveovarian cancer in late relapse. J Clin Oncol 2010;28(20):3323–9.http://dx.doi.org/10.1200/JCO.2009.25.7519.

[24] Armstrong DK, Bundy B, Wenzel L, et al. Intraperitonealcisplatin and paclitaxel in ovarian cancer. N Engl J Med2006;354(1):34–43. http://dx.doi.org/10.1056/NEJMoa052985.

[25] Wenzel LB, Huang HQ, Armstrong DK, Walker JL, Cella D.Health-related quality of life during and after intraperitonealversus intravenous chemotherapy for optimally debulked ovariancancer: a Gynecologic Oncology Group Study. J Clin Oncol2007;25(4):437–43. http://dx.doi.org/10.1200/JCO.2006.07.3494.

[26] von Gruenigen VE, Huang HQ, Gil KM, Frasure HE, ArmstrongDK, Wenzel LB. The association between quality of life domainsand overall survival in ovarian cancer patients during adjuvantchemotherapy: a Gynecologic Oncology Group Study. GynecolOncol 2012;124(3):379–82. http://dx.doi.org/10.1016/j.ygyno.2011.11.032, pii: S0090-8258(11)00950-4.

[27] Krivak TC, Tian C, Rose GS, Armstrong DK, Maxwell GL. AGynecologic Oncology Group Study of serum CA-125 levels inpatients with stage III optimally debulked ovarian cancer treatedwith intraperitoneal compared to intravenous chemotherapy: ananalysis of patients enrolled in GOG 172. Gynecol Oncol2009;115(1):81–5. http://dx.doi.org/10.1016/j.ygyno.2009.06.021,pii: S0090-8258(09)00459-4.

[28] Walker JL, Piedmonte MR, Spirtos NM, et al. Laparoscopycompared with laparotomy for comprehensive surgical staging ofuterine cancer: Gynecologic Oncology Group Study LAP2. J ClinOncol 2009;27(32):5331–6. http://dx.doi.org/10.1200/JCO.2009.22.3248.

[29] Kornblith AB, Huang HQ, Walker JL, Spirtos NM, RotmenschJ, Cella D. Quality of life of patients with endometrial cancerundergoing laparoscopic international federation of gynecologyand obstetrics staging compared with laparotomy: a GynecologicOncology Group study. J Clin Oncol 2009;27(32):5337–42. http://dx.doi.org/10.1200/JCO.2009.22.3529.

[30] Walker JL, Piedmonte MR, Spirtos NM, et al. Recurrence andsurvival after random assignment to laparoscopy versus laparot-omy for comprehensive surgical staging of uterine cancer:Gynecologic Oncology Group LAP2 Study. J Clin Oncol2012;30(7):695–700. http://dx.doi.org/10.1200/JCO.2011.38.8645.

[31] Osoba D. Translating the science of patient-reported outcomesassessment into clinical practice. J Natl Cancer Inst Monogr2007;37:5–11. http://dx.doi.org/10.1093/jncimonographs/lgm002.

[32] Greimel E, Kristensen GB, van der Burg ME, et al. Quality of lifeof advanced ovarian cancer patients in the randomized phase IIIstudy comparing primary debulking surgery versus neo-adjuvantchemotherapy. Gynecol Oncol 2013;131(2):437–44. http://dx.doi.org/10.1016/j.ygyno.2013.08.014, pii: S0090-8258(13)01099-8.

[33] Vergote I, Trope CG, Amant F, et al. Neoadjuvant chemotherapyor primary surgery in stage IIIC or IV ovarian cancer. N Engl JMed 2010;363(10):943–53. http://dx.doi.org/10.1056/NEJMoa0908806.