Four periods were defined: 21 … Within these four periods, changes in pain and physical well-being were weakly correlated with the maximum CA 19-9 decrease, more prominent after best CA 19-9 response as compared to before (e.g., pain <21 days Calcitriol before: R=0.27, P<0.05; <21 days after: R=0.34, P<0.005). To estimate whether this correlation was depending on the initial tumour load, patients were grouped according to the median CA 19-9 concentration at baseline. Only patients with a value below the median (27 �� ULN) showed a significant correlation between maximum CA 19-9 decrease and changes in pain within 20 days before (R=0.37, P<0.05). These patients reported better pain scores at baseline (median: 89.5 vs 67.0; P<0.05). In summary, baseline CA 19-9 had no relevant prognostic impact on QOL until treatment failure.

There was an association between the maximum CA 19-9 decrease and pain and physical well-being, respectively, but the maximum decrease was not prognostic for these domains: These group differences in pain and physical well-being were present already before the maximum decrease. Discussion We investigated the prognostic value of QOL relative to CA 19-9, and the role of CA 19-9 in estimating palliation in patients with advanced pancreatic cancer receiving chemotherapy within a randomised controlled clinical trial (Herrmann et al, 2007; Bernhard et al, 2008). At baseline, less pain and tiredness, that is, less symptom burden, predicted better survival, as shown for various QOL domains in other cancer sites (Gotay et al, 2008; Montazeri, 2009; Quinten et al, 2009).

In advanced disease, mainly physical functioning and symptoms have been prognostic. Pre-treatment fatigue was a dominant prognostic factor in patients with advanced head and neck carcinoma treated with radiotherapy (Fang et al, 2004). Similarly, pain and dysphagia were strong prognostic factors in patients with non-small-cell lung cancer (Efficace et al, 2006). We did not assess further patient-rated symptoms in our trial. Pain and tiredness showed consistent findings in all models. Their nonlinear association with survival may be confounded by individual analgesic treatment before the baseline assessment. These symptoms may serve as baseline characteristics for patients with advanced pancreatic cancer treated in clinical trials.

Their relative contribution Anacetrapib is, however, smaller than the effect of pre-treatment concentration of CA 19-9 on survival. In contrast, baseline CA 19-9 did not predict QOL or time on study treatment, besides a marginal effect on pain. Neither CA 19-9 nor QOL predicted tumour response to chemotherapy. Survival is influenced by different factors than response to chemotherapy, although response impacts on survival. Thus, CA 19-9 and QOL at baseline provide limited information for estimating palliation by chemotherapy. In patients with a better health status, QOL during chemotherapy differed by the timing relative to the best CA 19-9 response.