Results
Baseline patients’ characteristics are shown in Table 1. Among 168
patients, 108 patients (64%) had completed the 6 scheduled
administrations, 48 patients (29%) had discontinued
223Ra because of progressive disease or death, while
12 (7%) were still receiving therapy at the time of the analysis. A
total of 2264 questionnaires have been collected and analyzed, 1132 of
which were EORTC QLQ-C30 and 1132 were QLQ-BM22. The median follow-up
time from the first 223Ra-treatment to either death or
last contact (last 223Ra administration, last
follow-up phone call or last follow-up 99mTc-diphosphonate bone scan),
was 11±8 months (range 1-38). Median OS time was 12 months (95%CI 10 -
13 months), as shown in Figure 1. The univariate analysis evaluating the
prognostic value of all baseline clinical variables showed that
patients’ Hb, PLT, tALP, and PSA values were independently associated
with an increased risk of death. As shown in Table 2, almost all items
of both baseline EORTC QLQ-C30 and QLQ-BM22 questionnaires were
significantly associated with OS on univariate analysis. Only age
(p=0.095), dyspnea (p=0.511), diarrhea (p=0.055) and financial
difficulties (p=0.218) were not significantly associated with
improvement in OS. When adjusting for other measures on multivariate
analysis, baseline patients’ Hb, tALP, and two EORTC QLQ-C30 items
(PF2-physical functioning and DY-dyspnea) were significantly associated
with OS. After data reduction, the first principal component (PC)
explained 90% of the total variation, being then a satisfactory summary
of the nineteen questionnaire items. The weights of the first PC are
presented in Table 3. Each subject QoL score was calculated as the sum
of the product of each loading and the corresponding item measurement.
The score ranges from -222 to 200. The new QoL score showed good area
under the curve (AUC 0.73); ROC curve is shown in Figure 2. The
threshold selected by maximizing the sensitivity and specificity was 7.
The entire cohort was stratified into two subgroups on the basis of the
baseline QoL cut-off so obtained; the estimated overall survival and
log-rank test show that patients with a baseline QoL < 7 had a
median OS time of 8 months (95%CI 6 - 11 months), while those with a
baseline QoL ≥ 7 had a median OS time of 16 months (95%CI 12 - 24
months), showing a significant association between higher levels of
baseline QoL and longer survival (log-rank p<0.001) (Figure
3). The baseline value of our QoL score was significantly associated
with OS at univariate analysis (HR = 0.993, 95% CI 0.991-0.996) with a
p value <0.001 and when adjusting for other measures on
multivariate analysis, baseline patients’ Hb (HR = 0.816, 95% CI 0.717
– 0.927), tALP (HR = 1.008, 95% CI 1.003 – 1.013) and our QoL score
(HR = 0.995, 95% CI 0.992 – 0.998) were significantly associated with
OS (Table 4). The joint model showed a significant deterioration of
global-QoL during both 223Ra-therapy, and the
follow-up period (p < 0.001), as shown in Figure 4.