Quality of life as a prognostic factor of overall survival in patients
with advanced hepatocellular carcinoma: results from two French
Franck Bonnetain Æ Æ Xavier Paoletti Æ Æ Sandra Collette Æ Æ
Michel Doffoel Æ Æ Olivia Bouche ´ Æ Æ Jean Luc Raoul Æ Æ
Philippe Rougier Æ Æ Fadil Masskouri Æ Æ Jean Claude Barbare Æ Æ
Accepted: 28 May 2008
? Springer Science+Business Media B.V. 2008
(QoL) as a prognostic factor of overall survival (OS) and to
determine whether QoL data improved three prognostic
classifications among French patients with advanced
hepatocellular carcinoma (HCC).
We pooled two randomized clinical trials con-
ducted by the Fe ´de ´ration Francophone de Cance ´rologie
Digestive in a palliative setting. In each trial QoL was
assessed at baseline using the Spitzer QoL Index (0–10).
The aims of our study were to assess quality of life
Three prognostic classifications were calculated: Okuda,
Cancer of the Liver Italian Program (CLIP), and Barcelona
Clinic Liver Cancer group (BCLC) scores. To explore
whether the scores could be improved by including QoL,
univariate Cox analyses of all potential baseline predictors
were performed. A final multivariate Cox model was
constructed including only significant multivariate baseline
variables likely to result in improvement of each scoring
system. In order to retain the best prognostic variable to
add for each score, we compared Akaike information cri-
terion, likelihood ratio, and Harrell’s C-index. Cox
analyses were stratified for each trial.
Among 538 included patients, QoL at baseline was
available for 489 patients (90%). Longer median OS was
ranging from 2.17 months (Spitzer = 3) to 8.93 months
(Spitzer = 10). Variables retained in the multivariate Cox
model were: jaundice, hepatomegaly, hepatalgia, portal
thrombosis, alphafetoprotein, bilirubin, albumin, small HCC,
and Spitzer QoL Index (hazard ratio = 0.84 95% CI [0.79–
0.90]). According to Harrell’s C-index, QoL was the best
prognostic variable to add. CLIP plus the Spitzer QoL Index
had the most discriminating value (C = 0.71).
prognostic factor for survival in HCC patients with mainly
alcoholic cirrhosis.Theprognostic value ofCLIPscore could
be improved by adding Spitzer QOL Index scores.
Prognostic factor ? Overall survival ? Validation
Quality of life ? Hepatocellular carcinoma ?
Primaryliver cancer is the fifth most frequent cancerand the
third most common cause of cancer-related death in the
F. Bonnetain ? F. Masskouri
Methodological and Biostatistical unit, Fe ´de ´ration Francophone
de Cance ´rologie Digestive, INSERM U866, Dijon, France
F. Bonnetain (&)
Centre Georges Franc ¸ois Leclerc, 1 rue Professeur Marion,
BP 77980, 21079 Dijon cedex, France
X. Paoletti ? S. Collette
Institut National du Cancer, Paris, France
CHU Strasbourg, Strasbourg cedex, France
O. Bouche ´
Service d’He ´patogastroente ´rologie, Centre Hospitalo-
Universitaire R. Debre ´, Reims, France
J. L. Raoul
Centre Euge `ne Marquis, Rennes, France
P. Rougier ? L. Bedenne
Fe ´de ´ration Francophone de Cance ´rologie Digestive, INSERM
U866, Dijon, France
J. C. Barbare
Division a ` la recherche Clinique, CHU Amiens Nord, Amiens,
Qual Life Res
of liver cancer; this cancer generally develops following
cirrhosis or hepatitis B (HBV) or C (HCV) infections. The
incidence of HCC has substantially increased in developed
deaths per year are due to this cancer, whose main aetiology
is related to alcohol abuse.
Quality of life (QoL) is a major aspect in the care of
cancer patients, and has been recognized as an important
end point in cancer clinical trials and clinical practice,
along with the traditional end points including tumor
response rate, disease-free survival, and overall survival
[4–8]. More recently, pretreatment QoL has been recog-
nized as a potential prognostic factor of survival in cancer
patients [8–9]. Classification of patients according to their
prognosis is a central issue since inclusion criteria in
clinical trials supposes that homogenous groups of patients
can be identified. Various prognostic factors of overall
survival have thus been explored and several classifications
have been proposed for patients with HCC [10–13]. The
most commonly used scores are Okuda Stage, Cancer of
the Liver Italian Program (CLIP), Barcelona Clinic Liver
Cancer group (BCLC), and Groupe d’Etude et de Traite-
ment du Carcinome He ´patocellulaire (GRETCH). Different
studies have compared and ranked these classifications
[14–22] according to their prognostic value. The results of
the different studies have been discordant and remain
controversial. Furthermore, most of the studies focused on
HBV/HCV-infected patients, even though it is very likely
that overall survival depends on the aetiology of the cir-
rhosis. Therefore, the conclusions of these studies may not
be consistent with those based on alcohol-related HCC.
This study focuses on patients with advanced-stage
HCC mainly associated with alcoholic cirrhosis. Based on
a pooled analysis of two randomized clinical trials (RCT)
carried out by the Fe ´de ´ration Francophone de Cance ´rologie
Digestive (FFCD), we have assessed the value of quality-
of-life scores for predicting overall survival. We have also
explored whether staging systems for HCC could be
improved by adding quality-of-life data.
Patients and methods
We performed a pooled analysis of two RCTs of patients
with HCC in a palliative setting.
The FFCD 9403 trial evaluated the survival benefit of
adding tamoxifen to best supportive care. In this trial, 420
eligible patients from 78 French institutions were ran-
domized . Inclusion criteria were HCC not eligible for
surgical resection, liver transplantation, percutaneous
ablation, or transarterial chemoembolization. Diagnosis of
HCC was either cytologically or histologically confirmed,
or made by the association of a diagnosis of cirrhosis:
tomography (CT) scan, and/or an manetic resonance
imaging (MRI) showing a space-occupying lesion having
an image consistent with the diagnosis of HCC and per-
sistent alphafetoprotein (AFP) values above 500 lg/l.
Exclusion criteria were renal failure (serum creatinine
[130 lmol/l), advanced liver disease (Child-Pugh class
C), World Health Organization (WHO) performance status
(PS) greater than 2, and prior treatment with tamoxifen.
The FFCD 9402 trial evaluated the survival benefit of
alone. In this trial, 122 eligible patients from 15 French
institutions were randomly assigned . Inclusion criteria
were HCC not eligible for surgical resection, liver transplan-
tation or percutaneous ablation. All patients were cirrhotic
(cirrhosis diagnosis was histologically proven or based on
clinical and biological parameters). Diagnosis of HCC was
based on biopsy, or persistently elevated AFP levels
([400 lg/l) with one typical imaging finding (ultrasono-
graphy or CT scan or MRI, or normal AFP levels with 2
concordant imaging findings). Exclusion criteria were
advanced liver disease (Child-Pugh class C), advanced HCC
(Okuda stage III), portal vein thrombosis (trunk and primary
branches) orarteriovenous shunting, extrahepaticmetastases,
clearance \80 ml/min), platelet count \50 9 109/l, pro-
thrombinactivity\50%,and cardiac ejectionfraction\35%.
These two trials failed to demonstrate any superiority of
the investigated treatments [23–24].
In the 9403 trial, four patients, for whom more than 60%
of the data were missing, were excluded and in the 9402
trial, one patient, who had a WHO PS of 4, was excluded.
Finally, 122 patients in the 9402 trial and 416 patients in
the 9403 trial were retained and pooled (n = 538).
We further selected patients who had completed the
quality-of-life questionnaire at baseline, that is before
Quality of life (QoL) was evaluated before randomization
by the Spitzer QoL Index [25–27], which is a global can-
cer-specific QoL score. A score of 0 (worst) to 10 (best)
was calculated following the assessment of five dimensions
related to activity, daily life, health perceptions, social
support, and behavior. Each area was assessed by one item
rated on a three-point Likert scale. The QoL questionnaire
was completed by the patients in the two trials [23–24].
However, in the 9402 trial, to prevent missing QoL data,
when patients were unable to complete the questionnaire
Qual Life Res
due to progression of the cancer and/or poor health status,
clinicians were allowed to assess QoL on their behalf.
Table 1 presents definitions of Okuda, CLIP, and BCLC
prognostic scores. Furthermore, the Child-Pugh score,
required to calculate CLIP, was based on ascites, enceph-
alopathy, total bilirubin, prothrombin rate, and albumin.
Collected variables and reconciliation
The following baseline variables were retained to calculate
the prognostic classification, to explore their prognostic
value, and to determine whether they could improve stag-
ing systems: Spitzer QoL Index, age, sex, date and
modality of HCC diagnosis, date of death or of last
information on vital status, presence of cirrhosis and its
aetiology, clinical parameters (weight, oedema of the lower
limbs, jaundice, hepatomegaly,
encephalopathy), serological parameters (total bilirubin,
prothrombin rate, creatinine, albumin, AFP serum levels),
tumor characteristics (site of the principal tumor, maxi-
mum tumor diameter, number of tumor sites in the liver,
tumor extension, portal vein thrombosis, extrahepatic
metastases), and WHO performance status.
Biological parameters were dichotomized according to
usual reports in the literature and age according to the
Portal vein thrombosis in the two trials was reported
according to different criteria. The data were reconciled by
the principal investigators.
, that is, 1 nodule\50 mm or 2–3 nodules\30 mm.
Table 1 Definitions of the Okuda, CLIP, and BCLC classifications
Tumor morphologyUninodular and
Portal vein thrombosisNo
A1 A2 A3 A4BCD
Tumor stageSingle SingleSingle 3 tumors
Liver functional statusNo portal
Okuda stages: I = 0 points; II = 1–2 points; III = 3–4 points
CLIP: Cancer of the Liver Italian Program scoring system
BCLC: Barcelona Clinic Liver Cancer staging classification; stages A and B all criteria should be fulfilled; stage C at least one criterion PST 1–2
or vascular invasion or extrahepatic spread; stage D at least one criterion PST 3–4 or Okuda stage III/Child-Pugh C
Qual Life Res
All statistical analyses were performed on the pooled data
base stratified by trial to take into account differences
between the trials. Per-trial analyses were then performed
and these enabled us to check the robustness of our results.
Baseline variables are given as means (standard devia-
tion, SD), or frequencies and percentages. The Spitzer QoL
Index is shown as mean (SD) and median (minimum–
maximum) and the results for the two trials were compared
using the Mann–Whitney test.
Overall survival is defined as the time between the date
of inclusion and the date of death (all causes) or the date of
the last follow-up for living patients. Survival was esti-
mated using the Kaplan–Meier
compared using the stratified log-rank test. Median survival
was calculated with its 95% confidence interval (CI).
For prognosis purposes, the Spitzer QoL Index was
treated as a continuous ordered variable. However, to
represent survival graphically, we divided the Spitzer Index
into three subgroups (0–7 versus 8 versus 9 and 10).
Monotonicity of the gradients according to the Spitzer
QoL Index score was checked by comparing median sur-
vival times. Patients with a better prognosis should have
higher median values than patients with a poor prognosis.
A significant log-rank for trend was considered to reflect
Univariate and multivariate Cox analyses stratified by
trial were performed to estimate hazard ratios (HR) and its
95% confidence intervals (95% CI). We performed uni-
variate Cox analyses of all potential baseline predictors
including the variables constituting each score. We tested a
multivariate model including all variables with univariate
P\0.10, including bilirubin, which had a P-value close to
0.10. The final multivariate model was constructed with a
backward procedure among these variables to select vari-
ables likely to improve each scoring system. Internal
validity of this model has been explored using bootstrap-
ping (100 replications).
Finally, multivariate Cox model analyses were per-
formed for each score. The best models were built with
forward and backward procedures among baseline vari-
ables pertinent to improve each score. In order to retain the
best prognostic variable to add to each score, from the final
model we compared the Akaike information criterion
(AIC), the likelihood ratio (LR), and Harrell’s C statistic
. A smaller AIC value or a higher LR indicated that the
model was more informative regarding the prognosis of
overall survival. Harrell’s C statistic estimates the pro-
portion of correct predictions, i.e., the proportion of
patients with a better prognostic stage who have better
survival. Bootstrapping (100 replications) was applied
for internal validity to calculate optimism-corrected
C-statistics. The results of Harrell’s C-index varied from
0.5 (no discrimination) to 1 (perfect discrimination).
Harrell’s C-index was also calculated for the Spitzer
QoL Index score alone.
All data analyses were performed using Stata V10. A
P-value less than 0.05 was considered significant.
Among the 538 patients of the pooled database, 489 patients
had available QoL scores at baseline: 93 (76%) in the 9402
trial and 396 (95%) in the 9403 trial. Their baseline clinical
characteristics were similar to those of the whole population
(Table 2). In the 9402 trial all QoL questionnaires were
completed by the patients while clinicians were allowed to
assess QoL on behalf of the patients when they were unable
to complete the questionnaire.
Patients’ baseline characteristics are described in
Table 2. Males were predominant (88%), as were patients
aged C65 years (63%). All patients in the 9402 trial were
cirrhotic (inclusion criteria), and 91% patients of the 9403
trial were cirrhotic. Among them, 454 patients (78%) had
alcoholic cirrhosis. WHO PS 0 was more frequent (50.3%).
Finally, patients in the 9402 trial had a better clinical,
biological, and tumor status (Table 2). Due to the inclusion
criteria, the majority of patients were Child-Pugh class A
or B, Okuda I and II, CLIP 1–3, and BCLC B or C.
The mean QoL at baseline differed significantly
(Wilcoxon P B 0.0001) by trial; it was 8.6 (SD 1.3) and
7.6 (SD 1.8) in the 9402 and 9403 trial, respectively,
resulting in a clinical difference of 10% in the theoretical
score range. A majority of patients had a Spitzer score
between 7 and 10.
At the time the databases were closed, 459 (94%) patients
had died and only 30 patients (6%) were alive. The median
survival was 5.26 months (95% CI: 4.4–6.0).
Overall survival differed significantly by trial (log-rank
test: P\0.0001) and thus required stratified analyses
(Table 3). Median survival was longer in the 9402 trial:
13 months (8.2–16.8) versus 4.3 months (3.8–5.0).
According to the Spitzer QoL Index, median overall
survival varied significantly from 2.17 for a Spitzer 3 to
8.93 months for a Spitzer 10 (log-rank test for ordered
groups: P\0.0001) (Table 3). Figure 1 shows survival
curves according to the Spitzer score subgroups.
Harrell’s C statistic, which reflects discriminatory
capability, was 0.63 for the Spitzer QoL Index alone.
Qual Life Res
Table 2 Baseline characteristics of patients with HCC with or without available QoL data
Patients with available QoL dataAll patients
9402 9403Total Total
N = 489%
N = 538%
Spitzer QoL Index
00 0.001 0.251 0.20
10 0.000 0.000 0.00
20 0.001 0.251 0.20
30 0.008 2.028 1.64
40 0.00 194.8019 3.89
51 1.0827 6.8228 5.73
66 6.454611.62 52 10.63
7 12 12.9058 14.6570 14.31
8 16 17.209423.74 110 22.49
9 3234.4182 20.71114 23.31
102627.96 60 15.158617.59
Male81 87.10354 89.39 43588.96478 88.85
C65 45 48.3926165.91 306 62.58337 62.64
Present93 100.00 361 91.16454 92.84 49892.57
Yes 7580.6530777.53 382 78.12414 76.95
Yes6 6.45 77 19.4483 16.97 93 17.29
Yes 6165.5930777.53368 75.2640174.54
Yes1920.43103 26.01122 24.9513124.35
Involved liver volume
Portal vein thrombosis
Alpha-fetoprotein serum level (lg/l)
Total bilirubin (lmol/l)
4346.2422356.3126654.40 296 55.02
Prothrombin rate (%)
0 3739.78 7518.9411222.9012322.86
Qual Life Res
Analyses of prognostic factors
Univariate Cox analyses stratified by trial showed that the
following variables were significantly associated with
lower overall survival (Table 4): alcoholic cirrhosis, jaun-
dice, hepatomegaly, hepatalgia, ascites, involved liver
volume greater than 50%, tumor localization, portal vein
thrombosis, AFP serum level C200 lg/l, total bilirubin
C20 lmol/l, and WHO PS[0. An increase of one unit of
the Spitzer QoL Index was significantly associated with
longer survival [HR = 0.81 (0.77–0.86)]. Likewise, albu-
min C35 g/l, prothrombin activity C80%, and small HCC
In multivariate analysis including the above variables
and those with univariate P B 0.10 (age, other cirrhosis
aetiology, and oedema of the lower limbs) the following
remained significant independent baseline predictors:
jaundice, hepatomegaly, hepatalgia, ascites, portal vein
thrombosis, AFP level, albumin level, small HCC, and
Spitzer QoL Index.
The final multivariate model was constructed with a
backward procedure based on these variables plus total bili-
multivariate model retained the following significant prog-
nostic factors (Table 4): jaundice, hepatomegaly, hepatalgia,
HCC, and Spitzer QoL Index. Internal validity of this model
assessed by bootstrapping showed the following 95% CI:
jaundice ([0.93–1.99]; P = 0.108), hepatomegaly ([1.18–
1.91]; P = 0.001), hepatalgia ([1.09–1.89]; P = 0.011),
ascites (minimal vs. no [1.03–1.80]; P = 0.03, abundant vs.
no [0.65–2.17]; P = 0.571), portal thrombosis ([1.18–1.77];
P B 0.0001), AFP ([1.34–2.23]; P B 0.0001), total bilirubin
([0.97–1.62];P = 0.085),albumin([0.56–0.93];P = 0.011),
small HCC ([0.40–0.80]; P = 0.001), and Spitzer QoL Index
([0.80–0.90]; P B 0.0001).
The three scores investigated could thus be improved
with the following eligible variables which are not included
in the corresponding score (Table 5):
– For CLIP: jaundice, hepatalgia, hepatomegaly, and
Spitzer QoL Index.
For Okuda: hepatomegaly, hepatalgia, portal vein
thrombosis, AFP serum level, small HCC, and Spitzer
For BCLC: hepatomegaly, jaundice, hepatalgia, AFP
serum level, and Spitzer QoL Index.
Table 2 continued
Patients with available QoL dataAll patients
9402 9403 TotalTotal
N = 489%
N = 538%
1 50 53.7619649.49 24650.31276 51.30
26 6.45125 31.57131 26.79139 25.84
Child-Pugh A 67 72.04 208 52.5327556.24 30456.51
Child-Pugh B2627.96 17143.18197 40.29 21740.33
Child-Pugh C0 0.00 17 4.29 17 3.48173.16
I 6266.6713333.59 19539.88 22141.08
II 3032.26 22957.8325952.97279 51.86
III11.08 34 8.59 357.16 387.06
0 1010.75174.2927 5.52325.95
1 3234.41 8220.71 11423.31 12523.23
2 2931.18 10927.53 13828.22 155 28.81
31617.20 10727.0212325.15 13224.54
466.45 5914.90 6513.29 7213.38
5–600.00 225.56 224.50224.09
A10 10.753 0.76 13 2.66 142.60
B 2021.51 4210.61 6212.68 6812.64
C 6266.6731078.28 37276.07 41176.39
D1 1.08 4110.35 428.59458.36
Qual Life Res
Internal validity of these models assessed by boot-
strapping showed the following 95% CI:
– For CLIP: jaundice ([1.13–2.16], P = 0.007]), hepa-
talgia ([0.99–1.90]; P = 0.061), hepatomegaly ([0.99–
1.59]; P = 0.059), and Spitzer QoL Index ([0.80–0.91];
P B 0.0001).
– For Okuda: hepatomegaly ([1.18–1.82]; P = 0.001),
hepatalgia ([1.04–1.82]; P = 0.026), portal vein throm-
bosis ([1.09–1.84]; P = 0.008), AFP serum level
([1.35–2.16]; P B 0.0001), small HCC ([0.46–0.91];
P = 0.013), and Spitzer QoL Index ([0.80–0.90]; P B
Table 3 Overall survival related to the staging systems, Spitzer QoL Index, and WHO PS at inclusion (trial stratification)
P value Median (months)95% CI 1 year (%) 2 years (%)3 years (%)
9402 12.97[8.37;16.83]0.530.27 0.12
9403 4.33[3.77;5.03] 0.220.080.03
Spitzer QoL Index165.30
0 1.23[;] 0.000.00 0.00
1 [;] 0.000.00 0.00
2 0.33[;] 0.00 0.000.00
3 2.17 [0.60;5.90]0.00 0.000.00
4 2.27 [1.33;3.07]10.530.000.00
5 2.27 [1.47;2.60]3.73 0.00 0.00
6 3.30[2.40;5.17]17.09 4.27 4.27
7 3.10 [2.43;4.37]22.86 6.861.71
85.60 [4.27;7.90]29.266.57 3.29
9 7.67 [5.80;10.50]36.50 19.97 7.71
10 8.93 [6.77;11.73]38.06 22.64 10.26
WHO PS 60.86
-0 11.53[8.67;16.13] 49.6221.50 12.51
-1 4.73[4.10;5.80]24.74 11.593.52
-2 2.43[2.13;3.07]14.04 3.901.56
Child-Pugh A 7.50[5.90;8.93]35.1715.197.64
Child-Pugh B 3.43[2.60;4.23]18.78 7.842.10
Child-Pugh C 1.57[0.37;2.87]5.880.00 0.00
Okuda I11.40 [8.53;14.00]46.5921.8910.30
Okuda II 4.00[3.43;4.97]16.86 5.471.82
Okuda III1.43 [0.90;1.80]0.00 0.00 0.00
CLIP 0 23.03[16.23;26.17]77.7842.9912.90
CLIP 112.97[11.10;17.97]54.92 28.1814.28
CLIP 24.33[3.70;5.73]15.223.30 1.65
CLIP 3 4.60[3.57;5.53]19.673.27 2.18
CLIP 42.13 [1.37;2.60]6.28 3.140.00
CLIP 5–61.73[1.10;2.50]4.55 4.550.00
BCLC A21.37[13.77;40.43]76.92 46.1530.77
BCLC B 16.10[11.37;18.80]60.7122.7111.65
BCLC C4.67 [4.10;5.57]23.1010.003.61
BCLC D 1.53[0.93;1.90] 2.380.00 0.00
Qual Life Res
– For BCLC: hepatomegaly ([1.24–2.03]; P B 0.0001),
jaundice ([0.95–1.93]; P = 0.092), hepatalgia ([1.09–
1.90]; P = 0.009), AFP serum Level ([1.48–2.21]; P B
P = 0.001).
AIC and LR statistics highlighted the fact that the
Spitzer QoL Index was the most informative variable to be
added to the CLIP and Okuda (Table 5). Prognostic
information of the BCLC could be improved by adding
AFP or the Spitzer QoL Index even though the BCLC
included WHO PS (Table 5).
According to Harrell’s C-index, the discriminating value
of CLIP plus the Spitzer QoL Index (C = 0.71) was better
than that of Okuda plus the Spitzer QoL Index (c = 0.69)
and BCLC (C = 0.68). Furthermore, the discriminatory
capability of Okuda (c = 0.64) and BCLC (C = 0.62)
alone were closer to QoL alone (c = 0.63) while, with a
Harrell’s C-index of 0.67, the discriminatory capability of
CLIP alone was best. The optimism-corrected C-statistics
and its 95% CI confirmed these results, as shown in
Our results highlighted the fact that QoL assessed by the
Spitzer Index was a strong and independent prognostic
factor of overall survival time for French patients with
advanced HCC following mainly alcoholic cirrhosis.
Furthermore, the Spitzer QoL Index was the most infor-
mative variable toadd in
discriminating power of the existing staging systems. After
adjusting for the prognostic score, the Spitzer QoL Index as
well as other variables remained associated with overall
survival, suggesting that prediction of the prognosis could
be improved. Nevertheless, patient-reported baseline QoL
provides additional prognostic information that supple-
ments traditional clinical factors, and should be considered
a complementary prognostic tool for survival in patients
with advanced HCC.
This positive correlation between QoL data and survival
time has already been reported in various cancer sites and
more specifically in advanced cancer [9, 30–32]. These
sites include the breast [33–35], lung [36–39], oesophagus
[40–42], head and neck , colon , malignant mela-
noma , multiple myeloma , ovary , and
malignant glioma . Even though few studies have been
carried out in HCC patients, our results are in agreement
with a recent study from Yeo et al. , which showed that
role and emotional functioning and appetite loss of QLQ-
C30 were associated with survival time. However, our
study is the first to assess the QoL score as a prognostic
factor in a population with mainly alcoholic HCC aetiol-
ogy, which is associated with older age at diagnosis, poor
living conditions, and other complications due to alcohol-
ism. The independent prognostic value of QoL for these
patients suggests that a better Spitzer QoL score reflects
better physical and emotional functioning (e.g., because of
certain personality characteristics and/or social circum-
stances) within a group of patients with similar disease
characteristics (advanced HCC and cirrhosis). In our
opinion one of the first therapeutic goal, in the aim to
improve overall survival, could be to preserve or improve
QoL by controlling impact of disease and maybe alcoholic
dependency on physical and emotional functioning.
The major strengths of our study are that 80% of the
QoL data were available at baseline. The clinical charac-
teristics of these patients are similar to those of the whole
population of our pooled randomized clinical trials, which
limits most potential selection biases. A high standard of
follow-up was applied, resulting in a minimal rate of loss to
follow-up, a large number of events, and adequate overall
statistical power. To complement the analyses of the
prognostic value of QoL data, we used statistical methods
order toimprove the
06 1218 24
Time in Months
30 364248 5460
Spitzer 0 to 7
Spitzer 9 and 10
06 1218 243036 4248 5460
Time in Months
Fig. 1 Overall survival according to the Spitzer QoL Index (four
subgroups) using Kaplan-Meier estimation, n = 489
Qual Life Res
to estimate the discrimination and the effect of adding QoL
information to existing prognostic scores. These analyses
were performed in order to investigate how QoL data, as a
complement to the widely used and validated staging sys-
tems (CLIP, Okuda, and BCLC), could help clinicians plan
clinical trials and select populations. Improving the scores
is a delicate challenge. Due to the specificity of HCC,
which generally develops with underlying cirrhosis,
clinicians have to take into account the gravity of the
hepatic disease, the extension of the tumor as well as the
general status and finally QoL of the patient. In our study,
the Spitzer QoL Index was associated with survival after
adjustment on the CLIP score, making it a good candidate
for the construction of a new score. Other variables of
interest that were not reported in previous studies [16, 17,
22, 23] include the presence of jaundice, hepatomegaly or
Table 4 Univariate and multivariate analyses of baseline prognostic factors (Cox model)
Univariate Cox Full multivariate Cox Final multivariate Cox
HR 95% CI
P valueHR 95% CI
P valueHR 95% CI
Sex(F vs. M)0.90 [0.67;1.21]0.4902
Weight (kg)[C65 (F) and C75 (M)
(yes vs. no)
Age in years 0.84 [0.69;1.02]0.07520.96 [0.78;1.18]0.6842
Alcoholic cirrhosis(yes vs. no) 1.27[1.01;1.59] 0.04101.20[0.93;1.53]0.1602
HCV cirrhosis(yes vs. no)0.97 [0.74;1.28]0.8329
HBV cirrhosis (yes vs. no)0.81 [0.53;1.24]0.3376
Other cirrhosis(yes vs. no)0.65 [0.42;1.03]0.0647
Jaundice (yes vs. no)2.02 [1.58;2.58]1.38 [1.04;1.82]0.02651.36 [1.04;1.79]0.0254
Hepatomegaly (yes vs. no)1.57 [1.27;1.95]1.41 [1.11;1.79]0.00551.50 [1.19;1.89] 0.0005
Oedemas of the
(yes vs. no)1.21[0.97;1.51] 0.90[0.71;1.14] 0.3693
Hepatalgia(yes vs. no) 1.64[1.32;2.03]
1.45 [1.14;1.84]0.0024 1.43[1.14;1.80] 0.0017
Ascites(minimal vs. no) 2.00[1.60;2.49] 1.41[1.11;1.79]0.0190 1.36[1.08;1.72]0.0325
(abundant vs. no) 1.94[1.37;2.75] 1.25[0.82;1.88]1.19 [0.80;1.76]
(left vs. right) 1.02[0.83;1.26]
(bilateral vs. right) 1.53 [1.07;2.18]1.14[0.78;1.67]
(yes vs. no) 1.59[1.29;1.97]
(yes vs. no)1.76[1.45;2.13]
\0.0001 1.40[1.14;1.72]0.0014 1.45[1.19;1.77] 0.0003
(C200 vs.\200) 1.91[1.58;2.32]
Albumin (g/l) 0.61[0.50;0.74]0.77[0.63;0.95]0.01620.73 [0.59;0.89] 0.0019
(yes vs. no)0.66[0.49;0.89]0.0061
0.59[0.43;0.81] 0.00100.56[0.42;0.77] 0.0002
(continuous) 0.81[0.77;0.86]0.87 [0.81;0.93] 0.00010.84 [0.79;0.90]
WHO PS(PS 1 vs. PS 0)1.64[1.29;2.09]
(PS 2 vs. PS 0)2.44 [1.85;3.23] 1.36 [0.97;1.91]
Qual Life Res
hepatalgia. However, these variables raise concerns due to
their dependence on the clinical examination. Likewise, it
is interesting to observe that QoL data seem to be have
more prognostic power and are more informative than
these parameters, and particularly when compared with
general performance status. As an example of the benefit,
Table 5 Evaluation of the independent contribution of retained baseline variables for each prognostic score
Multivariate Cox modelMultivariate Cox model
Trial stratificationTrial stratificationC optimism-
HR 95% CI
P value ModelAIC LR
CLIP(CLIP 1 vs. CLIP 0) 1.19[0.75;1.88]
\0.0001 Without covariates4412.65– 0.5
(CLIP 2 vs. CLIP 0) 2.82[1.79;4.45]CLIP 4305.14117.51 0.67 [0.64;0.69]
(CLIP 3 vs. CLIP 0) 2.51 [1.59;3.98]+Spitzer QoL Index 4270.42 154.22 0.71 [0.67;0.73]
(CLIP 4 vs. CLIP 0)4.22[2.58;6.88]+Jaundice4288.52 136.130.68 [0.65;0.70]
(CLIP 5–6 vs. CLIP 0) 4.36 [2.37;8.02]+Hepatalgia4296.06 128.59 0.69 [0.65;0.71]
Jaundice(yes vs. no) 1.56[1.21;2.02]0.0007+Hepatomegaly4297.99 126.660.69 [0.66;0.70]
Hepatomegaly (yes vs. No) 1.26 [1.00;1.58]0.0486 Full model4250.70 179.95 0.73 [0.70;0.75]
Hepatalgia (yes vs. no)1.37 [1.10;1.71]0.0057
-Spitzer QoL Index4276.98 151.670.71 [0.68;0.73]
Spitzer QoL IndexContinuous0.85 [0.80;0.90]-Jaundice 4259.33169.31 0.72 [0.69;0.74]
-Hepatalgia 4256.02172.63 0.72 [0.69;0.74]
-Hepatomegaly 4252.69175.960.72 [0.69;0.74]
OKUDA (Okuda II vs. Okuda I)1.59[1.29;1.97]
\0.0001 Without covariates 4412.65– 0.5
(Okuda III vs. Okuda I) 4.57[3.03;6.88]OKUDA4321.09 95.560.64 [0.61;0.66]
Hepatomegaly(yes vs. no) 1.47 [1.16;1.85] 0.0012+Spitzer QoL Index 4289.83128.820.69 [0.66;0.71]
Hepatalgia (yes vs. no)1.38[1.10;1.71] 0.0046+AFP 4293.86124.79 0.68 [0.65;0.69]
(yes vs. no)
(yes vs. no)
+Hepatalgia 4304.92 113.720.67 [0.64;0.68]
1.71[1.40;2.08]+Portal thrombosis4307.34111.30 0.67 [0.64;0.69]
0.65 [0.48;0.87]+Small HCC 4313.86 104.790.65 [0.62;0.67]
Spitzer QoL Index Continuous 0.85[0.80;0.90]+Hepatomegaly 4316.00102.650.66 [0.63;0.68]
Full model4228.56200.09 0.74 [0.70;0.75]
-Spitzer QoL Index 4257.67 168.980.72 [0.69;0.74]
-AFP 4254.57172.08 0.72 [0.69;0.74]
-Portal thrombosis 4238.04188.610.73 [0.70;0.74]
-Hepatomegaly4237.55 189.100.73 [0.70;0.74]
-Small HCC 4235.73190.920.73 [0.70;0.75]
-Hepatalgia4234.28 192.370.73 [0.70;0.75]
BCLC (BCLC B vs. BCLC A) 1.48[0.76;2.89]
\0.0001 Without covariates4412.65– 0.5
(BCLC C vs. BCLC A) 2.12[1.13;3.97] BCLC4347.61 71.040.62 [0.59;0.63]
(BCLC D vs. BCLC A) 5.21[2.53;10.72]+AFP4317.31 103.340.67 [0.64;0.69]
Jaundice(yes vs. no)1.36 [1.03;1.78]0.0285
+Spitzer QoL Index 4320.9299.730.68 [0.65;0.70]
Hepatomegaly(yes vs. no)1.58[1.26;1.99]+Hepatomegaly4332.1588.50 0.65 [0.63;0.67]
Spitzer QoL Index
(yes vs. no)
1.44[1.16;1.80]+Hepatalgia 4333.7086.950.64 [0.62;0.66]
1.81 [1.49;2.21]+Jaundice 4339.87 80.770.64 [0.61;0.66]
0.87 [0.82;0.92]Full model 4262.15 166.500.72 [0.69;0.74]
-AFP4295.03 131.610.70 [0.69;0.72]
-Spitzer QoL Index 4280.55146.100.71 [0.67;0.73]
-Hepatalgia4270.32 156.330.72 [0.69;0.74]
-Jaundice 4264.73 161.920.72 [0.69;0.74]
All statistics were calculated based on Cox regression with stratification per trial. LR, likelihood ratio; LR estimates loss of adjustment by
calculating the difference of the deviance between models with and without the variable. AIC, Akaike information criterion. A smaller AIC value
or a higher LR indicates that model is more informative regarding prognosis of overall survival. C, Harrell’s C-index varies from 0.5 (no
discrimination) to 1 (perfect discrimination). C optimism-corrected and its 95% CI calculated using bootstrapping (100 replications)
Qual Life Res
the prognostic information of the BCLC , which
includes performance status in staging systems, could also
be improved by adding QoL. First these results highlighted
that QoL and performance status covered different health
measurements, and secondly that QoL seems to be more
On the one hand, we have demonstrated that the Spitzer
QoL Index could be the most interesting data to include in
existing models to better predict overall survival among
patients with advanced HCC. On the other hand, the
prognostic classifications contain characteristics that are
part of the exclusion criteria of our trials. Variations in the
classification scores are therefore reduced, which leads to
lower discriminative power than that in the whole group of
advanced HCC patients. In such situations, new variables
such as QoL are more likely to improve the prognostic
ability of the classifications. However, it is well known that
the results of prognostic evaluations on the same data
overestimate the performance of any new prognostic score
or the prognostic value of QoL. As highlighted by Altman
and Royston  neither internal nor temporal evaluation
addresses the wider issue of the generalizability of the
model. The reproducibility of a prognostic model is defined
as the performance of a model on a sample of similar
patients not included in the development of the model. Our
results thus need to be validated in another trial. We plan to
perform this external validation on patients included in the
randomized FFCD CHOC trial investigating long-acting
octreodid treatment versus placebo in patients with
advanced HCC . Furthermore, this trial used multidi-
mensional European Organization for Research and
Treatment of Cancer (EORTC) QLQ-C30 QoL assessment,
which makes it possible to explore which QoL dimensions
would be predictive, and finally to confirm if general health
is predictive of overall survival.
One of the major limits of this pooled study is that QoL
was assessed using a cancer-specific global QoL tool. On
the one hand, we agree that multidimensional QoL would
be more informative than global QoL regarding which
parameters could predict overall survival . On the other
hand, in a setting of advanced HCC, it would be difficult
for patients to complete a 30-item questionnaire; the pro-
cess would be more time consuming and there would be a
higher rate of missing scores. As suggested by Lipscomb
et al., in some cases, a short simple (even single-question)
patient-reported evaluation of outcome is appropriate and
adequate . In this way, the Spitzer QoL Index could be
proposed as an acceptable alternative tool to prevent
missing data due to cancer progression and/or poor health
status. Furthermore, even though multidimensional mea-
sures are more informative [54–57], single-item or global
tools have already demonstrated their clinical values in
Since the investigated patients had mainly alcoholic
cirrhosis and were not eligible for curative treatments, they
were exclusively in a palliative setting. Even though
between 60% and 75% of all patients in France are treated
in this setting , the patients in this study formed a rather
homogeneous sample that was not representative of the
whole HCC population. Therefore, the conclusions are
limited to this specific population and cannot be extended
to less advanced patients. The effect of QoL on survival in
such patients would require a separate study.
Quality of life is a well-established end point for treat-
ment comparisons, and this study provides further reasons
for measuring QoL both in cancer research and clinical
practice for patients with advanced HCC. Further research
is needed to identify specific baseline QoL parameters that
are relevant to these patients. However, we could suggest
that global QoL scores should be components of all QoL
questionnaires used in phase III trials as they may be used
to measure the impact of treatments on patients’ well-being
as well as to predict prognosis. Lipscomb et al. 
underlined the fact that the use of QoL as an established
and accepted end point in cancer required the study of the
prognostic value of QoL. Studies using different QoL tools
and different cancer sites are necessary to confirm the value
of QoL in determining prognosis in cancer patients.
1. Parkin, D. M., Bray, F., Ferlay, J., & Pisani, P. (2001). Estimating
the world cancer burden. GLOBOCAN 2000. International
Journal of Cancer, 94, 153–156. doi:10.1002/ijc.1440.
2. El-Serag, H. B., Davila, J. A., Petersen, N. J., & McGlynn, K. A.
(2003). The continuing increase in the incidence of hepatocellular
carcinoma in the United States - An update. Annals of Internal
Medicine, 139, 817–823.
3. Remontet, L., Esteve, J., Bouvier, A. M., et al. (2003). Cancer
incidence and mortality in France over the period 1978–2000.
Revue d’Epidemiologie et de Sante Publique, 51, 3–30.
4. Garcia, S. F., Cella, D., Clauser, S., et al. (2007). Enhancing
patient-reported outcomes assessment in cancer clinical trials:
The PROMIS initiative. Journal of Clinical Oncology, 25, 5106–
5. Garcia, S. F., Cella, D., Clauser, S., et al. (1995). Editorial.
Quality of life and clinical trials. Lancet, 346, 1–2.
6. Beitz, J., Gnecco, C., & Justice, R. (1996). Quality-of-life end
points in cancer clinical trials: The U.S. Food and drug admin-
istration perspective. Journal of the National Cancer Institute.
Monographs, 20, 7–9.
7. Johnson, J. R., & Temple, R. (1985). Food and drug adminis-
tration requirements for approval of new anticancer drugs.
Cancer Treatment Reports, 69, 1155–1159.
8. Lipscomb, J., Donaldson, M. S., Arora, N. K., et al. (2004).
Cancer outcomes research. Journal of the National Cancer
Institute. Monographs, 33,
9. Dancey, J., Zee, B., Osoba, D., et al. (1997). Quality of life
scores: An independent prognostic variable in a general popula-
tion of cancer patients receiving chemotherapy. The National
Qual Life Res
Cancer Institute of Canada Clinical trials Group. Quality of Life
Research, 6, 151–158. doi:10.1023/A:1026442201191.
10. Okuda, K., Ohtsuki, T., Obata, H., et al. (1985). Natural history of
hepatocellular carcinoma and prognosis in relation to treatment.
Study of 850 patients. Cancer, 56, 918–928. 10.1002/1097-
11. The Cancer of Liver Italian Program Investigators. (1998). A new
prognostic system for hepatocellular carcinoma: A retrospective
study of 435 patients: The Cancer of the Liver Italian Program
(CLIP) investigators. Hepatology (Baltimore, Md.), 28, 751–755.
12. Llovet, J. M., Bru, C., & Bruix, J. (1999). Prognosis of hepato-
cellular carcinoma: The BCLC staging classification. Seminars in
Liver Disease, 19, 329–338.
13. Chevret, S., Trinchet, J. C., Mathieu, D., et al. (1999). A new
prognostic classification for predicting survival in patients with
hepatocellular carcinoma. Groupe d’Etude et de Traitement du
Carcinome Hepatocellulaire. Journal of Hepatology, 31, 133–
14. Ueno, S., Tanabe, G., Sako, K., et al. (2001). Discrimination
value of the new western prognostic system (CLIP score) for
hepatocellular carcinoma in 662 Japanese patients. Cancer of
Liver Italian Program. Hepatology (Baltimore, Md.), 34, 529–
15. Dilou, N., Patouillard, B., & Audigier, J. C. (2004). Les classi-
fications de pre ´diction de survie du carcinome he ´patocellulaire.
Gastroenterologie Clinique et Biologique, 28, 359–366. doi:
16. The Cancer of Liver Italian Program Investigators. (2000). Pro-
spective validation of the CLIP score: A new prognostic system
for patients with cirrhosis and hepatocellular carcinoma. Hepa-
tology (Baltimore, Md.), 31, 840–845. doi:10.1053/he.2000.5628.
17. Levy, I., Sherman, M., & The Liver Cancer Study Group of the
University of Toronto (2002). Staging of hepatocellular carci-
noma: Assessment of the CLIP, Okuda, and Child-Pugh staging
systems in a cohort of 257 patients in Toronto. Gut, 50, 881–885.
18. Wildi, S., Pestalozzi, B. C., McCormack, L., & Clavien, P. A.
(2004). Critical evaluation of the different staging systems for
hepatocellular carcinoma. British Journal of Surgery, 91, 400–
19. Farinati, F., Rinaldi, M., Gianni, S., & Naccarato, R. (2000). How
should patients with hepatocellular carcinoma be staged? Vali-
dation of a new prognostic system. Cancer, 89, 2266–2273.
20. Marrero, J. A., Fontana, R. J., Barrat, A., et al. (2005). Prognosis
of hepatocellular carcinoma: Comparison of 7 staging systems in
an American cohort. Hepatology (Baltimore, Md.), 41, 707–716.
21. Leung, T. W., Tang, A. M., Zee, B., et al. (2002). Construction of
the Chinese university prognostic index for hepatocellular car-
cinoma and comparison with the TNM staging system, the Okuda
staging system, and the cancer of the liver Italian program staging
system: A study based on 926 patients. Cancer, 94, 1760–1769.
22. Cillo, U., Vitale, A., Grigoletto, F., et al. (2006). Prospective
validation of the Barcelona clinic liver cancer staging system.
Journal of Hepatology, 44, 723–731. doi:10.1016/j.jhep.2005.
23. Barbare, J. C., Bouche ´, O., Bonnetain, F., et al. (2005). Ran-
domized controlled trial of tamoxifen in advanced hepatocellular
carcinoma. Journal of Clinical Oncology, 23, 4338–4346. doi:
24. Doffoe ¨l, M., Bonnetain, F., Bouche ´, O., et al. (2008). Multicentre
randomised phase III trial comparing tamoxifen alone or with
transarterial lipiodol chemoembolisation for unresectable hepa-
Francophone de Cance ´rologie Digestive 9402). European Jour-
nal of Cancer, 44, 528–538.
25. Spitzer, W. O., Dobson, A. J., Hall, J., et al. (1981). Measuring
the quality of life of cancer patients. A concise QL-Index for use
by physician. Journal of Chronic Diseases, 34, 585–597. doi:
26. Anderson, R. T., Aaronson, N. K., & Wilkin, D. (1993). Critical
review of the international assessments of heath-related quality of
life. Quality of Life Research, 2, 369–395. doi:10.1007/
27. Sloan, J. A., Loprinzi, C. L., Kuross, S. A., et al. (1998). Ran-
domized comparison of four tools measuring overall quality of
life in patients with advanced cancer. Journal of Clinical
Oncology, 16, 3662–3673.
28. Mazzaferro, V., Regalia, E., Doci, R., et al. (1996). Liver trans-
plantation for the treatment of small hepatocellular carcinomas in
patients with cirrhosis. New England Journal of Medicine, 334,
29. Harrell, F. E., Lee, K. L., & Mark, D. B. (1996). Tutorial in
biostatistics. Multivariable prognostic models issues in develop-
ing models, evaluating assumptions and adequacy, and measuring
and reducing errors. Statistics in Medicine, 15, 361–387.
30. Ringdal, G. I., Ringdal, K., Kvinnsland, S., et al. (1994). Quality
of life of cancer patients with different prognosis. Quality of Life
Research, 3, 143–154.
31. Ringdal, G. I., Gotestam, K. G., Kaasa, S., et al. (1996). Prog-
nostic factors and survival in a heterogeneous sample of cancer
patients. British Journal of Cancer, 73, 1594–1599.
32. Coates, A., Porzsolt, F., & Osoba, D. (1997). Quality of life in
oncology practice: Prognostic value of EORTC QLQ-C30 scores
in patients with advanced malignancy. European Journal of
Cancer, 33, 1025–1030.
33. Coates, A., Forbes, J., & Simes, R. J. (1993). Prognostic value of
performance status and quality of life scores during chemother-
apy for advanced breast cancer. The Australian New Zealand
Breast Cancer Trials Group. Journal of Clinical Oncology, 11,
34. Kramer, J. A., Curran, D., Piccart, M., et al. (2000). Identification
and interpretation of clinical and quality of life prognostic factors
for survival and response to treatment in first-line chemotherapy
in advanced breast cancer. European Journal of Cancer, 36,
35. Efficace, F., Biganzoli, L., Piccart, M., et al. (2004). Baseline
health-related quality of life data as prognostic factors in a phase
III multicentre study of women with metastatic breast cancer.
European Journal of Cancer, 40, 1021–1030.
36. Langendijk, H., Aaronson, N. K., de Jong, J. M. A., et al. (2000).
The prognostic impact of quality of life assessed with the EORTC
QLQ-C30 in inoperable non-small cell lung carcinoma treated
with radiotherapy. Radiotherapy and Oncology, 55, 19–25.
37. Montazeri, A., Milroy, R., Hole, D., et al. (2001). Quality of life
in lung cancer patients as an important prognostic factor. Lung
Cancer, 31, 233–240.
38. Herndon, J. E., Fleishman, S., Kornblith, A. N., et al. (1999). Is
quality of life predictive of the survival of patients with advanced
non small cell lung carcinoma? Cancer, 85, 333–340.
39. Dharma-Wardene, M., Au, H. J., Hanson, J., et al. (2004).
Baseline FACT-G score is a predictor of survival for advanced
lung cancer. Quality of Life Reseach, 13, 1209–1216.
40. Blazeby, J. M., Brookes, S. T., & Alderson, D. (2001). The
prognostic value of quality of life scores during treatment for
oesophageal cancer. Gut, 49, 227–230.
41. Fang, F. M., Tsai, W. L., Chiu, H., et al. (2004). Quality of life as
a survival predictor for esophageal squamous cell carcinoma
patients(Fe ´de ´ration
Qual Life Res
treated with radiotherapy. International Journal of Radiation,
Biology, Physics, 58, 1394–1404.
42. Chau, I., Norman, A. R., Cunningham, D., et al. (2004). Multi-
variate prognostic factor analysis in locally advanced and
metastatic esophago-gastric cancer-pooled analysis from three
multicenter, randomized, controlled trails using individual
patients data. Journal of Clinical Oncology, 22, 2395–2403.
43. de Graeff, A., de Leeuw, J. R. J., Ros, W. J. G., et al. (2001).
Sociodemographic factors and quality of life as prognostic indi-
cators in head and neck cancer. European Journal of Cancer, 37,
44. Ramsey, S. D., Anderson, M. R., Etzioni, R., et al. (2000).
Quality of life in survivors of colorectal carcinoma. Cancer, 39,
45. Coates, A., Thomson, D., McLeod, G. R., et al. (1993). Prog-
nostic value of quality of life scores in a trial of chemotherapy
with or without interferon in patients with metastatic malignant
melanoma. European Journal of Cancer, 29A, 1731–1734.
46. Wisloff, F., Hjorth, M., et al., for the Nordic Myeloma Study
Group. (1997). Health-related quality of life assessed before and
during chemotherapy predicts for survival in multiple myeloma.
British Journal of Haematology, 97, 29–37.
47. Kornblith, A. B., Thaler, H. T., Wong, G., et al. (1995). Quality
of life of women with ovarian cancer. Gynecologic oncology, 59,
48. Langendijk, H., Aaronson, N. K., ten Velde, G. P., et al. (1999).
Pretreatment quality of life in patients with gIioblastoma multi-
forme. Oncology Nursing Forum, 26, 921–925.
49. Yeo, W., Mo, F. K., Koh, J., et al. (2006). Quality of life is
predictive of survival in patients with unresectable hepatocellular
carcinoma. Annals of Oncology, 17, 1083–1089.
50. Altman, D., & Royston, P. (2000). What do we mean by vali-
dating a prognostic model? Statistics in Medicine, 19, 453–473.
51. Barbare, J. C., Bouche ´, O., Bonnetain, F., et al. (2005). Treatment
of advanced hepatocellular carcinoma with long-acting octreo-
tide: Preliminary results of a randomized placebo-controlled trial
(FFCD-ANGH 2001–01 CHOC). Journal of Clinical Oncology,
2005 ASCO Annual Meeting Proceedings, 23(16S), 4036.
52. Gill, T. M., & Feinstein, A. R. (1994). A critical appraisal of the
quality of quality-of-life measurements. The Journal of the
American Medical Association, 272, 619–626.
53. Lipscomb, J., Reeve, B. B., Clauser, S. B., et al. (2007). Patient-
reported outcomes assessment in cancer trials: Taking stock,
moving forward. Journal of Clinical Oncology, 25, 5133–5140.
54. Simes, R. J., Greatorex, V., & Gebski, V. J. (1998). Practical
approaches to minimize problems with missing quality of life
data. Statistics in Medicine, 17, 725–737.
55. Bernhard, J., Cella, D. F., Coates, A. S., et al. (1998). Missing
quality of life data in cancer clinical trials : Serious problems and
challenges. Statistics in Medicine, 17, 517–532.
56. Bernhard, J., Sullivan, M., Hurny, C., et al. (2001). Clinical rel-
evance of single item quality of life indicators in cancer clinical
trials. British Journal of Cancer, 84, 1156–1165.
57. Sloan, J. A., Aaronson, N., Cappelleri, J. C., et al., Clinical
Significance Consensus Meeting Group. (2002). Assessing the
clinical significance of single items relative to summated scores.
Mayo Clinic Proceedings, 77, 479–487.
58. Caumes, J. L., Nousbaum, J. B., Bessaguet, C., et al. (2007).
Epidemiology of hepatocellular carcinoma in Finiste `re. Pro-
spective study from June 2002 to May 2003. Gastroenterologie
Clinique et Biologique, 1, 259–264.
Qual Life Res