

S102
A B S T R A C T S
Methods:
A prospective observational, multicentre study
including patients
70 years with a newly diagnosed,
histologically confirmed cancer or first relapse after previous
curativetreatment.Atinclusion,themGAwasperformedpartly
by trained nurses and partly using self-report questionnaires.
The assessment included eight domains; nutrition (Patient-
Generated Subjective Global Assessment), comorbidity
(Physical Health Section, Older Americans’ Resources and
Services comorbidity scale), regular medications, falls the
past six months, activities of daily living, physical (Timed
Up and Go), cognitive (Mini Mental State Examination) and
emotional function (Geriatric Depression Scale-15). For each
domain, a cut-off value for impaired function was defined.
Patients were categorized as frail if at least one domain was
registered as impaired. All other patients were categorized as
non-frail. Blinded for the mGA categorization and without any
specific instructions, the oncologists were asked to categorize
patients as fit, intermediate or frail. The intermediate (n=89)
and frail (n=15) category was merged into one category
named “suspected frail”. The agreement between mGA frail
and “suspected frail” was assessed by Kappa statistics. The
association of frailty and ”suspected frailty” with OS was
investigated by bi- and multivariate Cox regression analyses
adjusting for age, sex, cancer diagnosis, ECOG Performance
Status, stage and treatment.
Results:
This abstract presents baseline data. Between
January 2013 and April 2015, 307 patients were enrolled at 8
hospitals. Of these, 288 patients (94%) completed all baseline
questionnaires and thus the mGA, 286 patients had frailty
rated by the oncologists. Median age was 77 (70-95) years,
160 patients (56%) had distant metastasis, most common
cancer diagnoses were colorectal (n=83, 29%), lung (n=59,
21%) and prostate (n=56, 19%). In all, 140/288 (49%) patients
were categorized as mGA-frail, 104/286 (36%) patients were
“suspected frail”. The overall agreement between the two
frailty categorizations was 65%, Kappa measurement value
was 0.30 (95% CI 0.19; 0.41), representing fair agreement. The
median follow up time for survival was 16.9 months; 45%
of patients were alive at date of censoring. In bivariate Cox
regression analyses being frail according to both the mGA (HR
1.86 (95% CI 1.36; 2.56)) and the oncologists’ clinical judgment
(HR 1.94 (95% CI 1.41; 2.66)) was significantly associated with
reduced OS (p<.01). In the multivariate models, only mGA
frailty was significantly associated with reduced OS with a HR
of 1.61 (95% CI 1.14; 2.27) (p<.01).
Conclusion:
A systematic mGA identified more patients
as frail than the oncologists’ subjective clinical judgment.
Frailty categorized by the modified geriatric assessment was
an independent negative prognostic factor and provides
important, additional prognostic information for oncologists.
Disclosure of interest:
None declared
Keywords:
Frailty, geriatric assessment
P112
A PROSPECTIVE NON-INTERVENTIONAL STUDY ON
THE USE OF BEVACIZUMAB AND CONVENTIONAL
CHEMOTHERAPY FOR FIRST-LINE TREATMENT OF
METASTATIC COLORECTAL (MCRC) PATIENTS: SCREENING
AND GERIATRIC ASSESSMENT (GA)
L. Decoster
1,
*, C. Kenis
2
, G. Houbiers
3
, B. Naessens
4
,
M. De Man
5
, G. Lambrecht
6
, V. Moons
7
, E. Monsaert
8
,
P. Vergauwe
9
, H. Prenen
2
, E. Van Cutsem
2
, E. Beutels
10
,
D. Frijns
11
, H. Wildiers
2
1
UZ Brussels, Brussels,
2
UZ Leuven, Leuven,
3
Gastroenterology
and Digestive Oncology, CHC Clinique St Joseph, Liege,
4
AZ
Nikolaas, St.Niklaas,
5
OLV Aalst, Gent,
6
AZ Damiaan, Oostende,
7
Imeldaziekenhuis, Bonheide,
8
AZ Maria Middelares, Gent,
9
AZ
Groeninge, Kortrijk,
10
Innosens,
11
Roche NV/SA, Brussels, Belgium
Introduction:
In Belgium, the average age at diagnosis of
CRC is 70.1 years for men and 71.6 years for women. Few
elderly with cancer are included in clinical trials and aging
is a heterogeneous process. Therefore, choosing the most
appropriate therapy and weighing the risks and benefits of
chemotherapy in older cancer patients is challenging. Geriatric
screening and GA allow evaluation of individual global health
status and treatment optimisation. This observational study
aims to complement the knowledge on chemotherapy and
bevacizumab usage in elderly with mCRC in current practice
in Belgium and evaluates the impact of baseline geriatric
screening and GA on treatment duration (TD), progression
free survival (PFS) and severe toxicity.
Objectives:
Primary objective was TD of first-line bevacizu-
mab-containing chemotherapy. Secondary objectives were
TD of conventional chemotherapy, safety of bevacizumab in
elderly and correlation of baseline geriatric screening and GA.
Methods:
252 patients
70 years with mCRC receiving
chemotherapy with/without bevacizumab were included in
the study. Geriatric screening with G8 and Flemish Triage
Risk Screening Tool (fTRST), Eastern Cooperative Oncology
Group (ECOG), as well as GA including activities of daily
living (ADL), instrumental activities of daily living (IADL),
Mini-Mental State Examination (MMSE), Geriatric Depression
Scale (GDS-15), Mini Nutritional Assessment (MNA), Charlson
Comorbidity Index (CCI), and Mobility-Tiredness Test (Mob-T)
was performed in all patients at baseline. Logrank tests (for
TD and PFS), Wilcoxon or Student t-tests (for severe toxicity)
and multivariate analyses were used for correlations with
the different screening and GA components. A subgroup
analysis excluding patients with ECOG
2 at baseline was
also performed.
Results:
In the total safety population, median TD (95%
CI) was 5.5 (5.1-6.2) months. The only baseline parameters
significantly associated with TD in univariate analysis
were ECOG
1, which was only 14.6% of patients, and MNA
(p=0.0006 and p=0.0162, respectively), while G8 showed a
trend (p=0.0607). Significant correlations were observed
for PFS versus ECOG (p<0.0001), MNA (p=0.0001) and G8
(p=0.0208) and for severe toxicity versus ECOG (p<0.0001) and
G8 (p=0.005). Both TD and PFS were significantly associated
with G8 (p=0.0093 and p=0.0002, respectively) when lowering
the G8 cut-off to 12 (i.e. median value). fTRST did not show