As previously described, a cohort of 75 patients was identified. PDAC diagnosis was based on histology/cytology in 63 patients (84%) and radiology with or without serum markers in 12 (16%) . Basic patient characteristics, treatment parameters and safety, as well as survival data in the overall population, have been published previously . A summary of survival data in the total population and subpopulations is presented in Table 1 and Table 2.
OS and PFS estimates for subgroups according to age, gender, ECOG performance status, tumour stage, serum CA19-9, serum albumin, serum bilirubin, myelosuppression, and occurrence of dose reduction, were made with Kaplan Meier survival analyses as seen in Figure 1 and Figure 2.
Univariable regression analyses
The results of the univariable Cox regression analyses with hazard ratios (HR) for OS and PFS are displayed in Table 1 and Table 2, revealing that serum albumin before treatment start (cut off 37g/l, HR 0.53, p=0.032), occurrence of thrombocytopenia (all CTCAE grades during treatment, HR 0.29, p=0.001), and bone marrow toxicity (CTCAE grade 3-4 during treatment, HR 0.41, p=0.012) were statistically significant prognostic markers predicting death (Table 1). A statistical trend, although not significant, was found between age (inverted correlation, HR 0.57, p=0.062) and OS.
Similarly, with regard to secondary endpoint PFS, univariable analyses revealed that the occurrence of thrombocytopenia of any CTCAE grade during treatment (HR 0.31, p=<0.001), treatment dose (HR 2.10, p=0.045), and bone marrow toxicity (CTCAE grade 2-4 during treatment, HR 0.50, p=0.021) were statistically significant for the prediction of progressive disease (Table 2).
Multivariable regression analyses
Parameters considered statistically relevant with regard to OS were further analysed in multivariable regression analyses; age >/≤ 65 years (inverted, with older age being associated with improved survival, HR 0.50, p=0.039), and S-Albumin (cut off 37g/l, HR 0.48, p=0.023), were statistically significant with regard to the prediction of death (Table1). A statistical trend, although not significant, was found between occurrence of thrombocytopenia (all CTCAE grades, HR 0.54, p=0.073) and OS.
Multivariable regression analyses with regard to the secondary endpoint PFS confirmed that thrombocytopenia of any CTCAE grade (HR 2.16, p=0.017) was an independent prognostic marker for progression (Table 2). A statistical trend, although not significant, was found between age (HR 0.62, p=0.107, with older age related to lower risk) and PFS. The inclusion of covariates in terms of occurrence of second line therapy and bone marrow toxicity of CTCAE grade 2-4 (instead of grade 3-4) did not affect the outcome of the multivariable analyses (data not shown).
Baseline characteristics and treatment data in ‘old’ vs. ‘young’ patients
As the finding of age above 65 years being a positive prognostic factor for OS was somewhat unexpected, further analyses of the characteristics of the age subgroups were made. While no statistically significant differences concerning baseline characteristics were evident in old and young patients, there was a trend towards more patients with metastatic disease (81% vs 62%, p=0.070), higher median CA19-9 (980 vs 475.5, p=0.057), and a history of less previous adjuvant chemotherapy (22% vs 41%, p=0.081) in patients ≤65. Dose intensity of NabP, but not Gem, was significantly lower in the ‘old’ subgroup (63% vs 75%, p=0.028, Table 3.)
Comparisons in patients with and without treatment associated thrombocytopenia
Further analyses were conducted in the subgroups of patients with and without signs of thrombocytopenia during the treatment course. Student’s t-test confirmed longer treatment duration in patients where thrombocytopenia occurred, with average number of treatment cycles of 6.3 and 3.4 in the thrombocytopenia versus the non-thrombocytopenia groups, respectively (p=0.003) (Table 4).
In addition, overall survival in the quartile of patients with the most pronounced reduction of blood platelet count at cycle 1 day 15 was compared with the quartile of patients with the least pronounced (or no) reduction of platelets at this time point, revealing no significant difference (Figure 3).