Correlation of skin rash and overall survival in patients with pancreatic cancer treated with gemcitabine and erlotinib – results from a non-interventional multi-center study

Between 2012 and 2015, 376 patients were enrolled across 85 German study sites. Out of these, 338 patients received at least one dose of erlotinib and constituted the safety analysis set (SAF). 270 patients commenced treatment with gemcitabine/erlotinib (according to therapy guidance) and comprised the full analysis set (FAS). The median duration from diagnosis of metastatic disease until informed consent was 27 days. Fifty-eight patients (21.5%) in the FAS had received previous palliative chemotherapy before treatment with gemcitabine/erlotinib potentially explaining the mean duration of 71 days from occurrence of metastatic disease until informed consent for this non-interventional study. The study flow is depicted in Fig. 1. Baseline demographics and disease characteristics are shown in Table 1 and Table 2.

Of note, non-M1 tumor stage was documented in 7% of the patients. Most probably this reflects the stage at the time of the initial diagnosis rather than tumor stage at inclusion into the study as use of erlotinib according to the summary of product characteristics (SmPC) was mandatory in this study (see Table 2). Due to the nature of this post-approval, non-interventional study, detailed follow-up of these cases was not feasible.

In the FAS population, 99 patients (36.7%) developed any skin rash, while 171 patients (63.3%) did not present with rash. Patients were stratified based on the appearance of a skin rash (dichotomous classification: rash-positive vs. rash-negative) and treatment effectiveness was calculated based on this stratification. The median overall survival (OS) of all individuals was 9.11 months with an OS of 9.93 months in rash-positive patients (n = 99; 36.7%) and 8.68 months in rash-negative patients (n = 171, 63.3%) (p = 0.2361, Log-Rank test) (Fig. 2a). There was no statistical difference in 1-year survival rates between the two groups (not shown). Preplanned subgroup analyses by rash grade revealed a trend towards better median OS in patients with rash grade ≥ 2 (n = 51, 13.91 months) when compared to patients with rash grade 1 (n = 48, 8.19 months) and patients without rash (n = 171, 8.68 months). However, this trend did not reach statistical significance (p = 0.0893, Log-Rank test), most likely due to the relatively small sample sizes (Additional file 1: Figure S1). Independent of skin reactions, there was a tendency of improved median OS in patients aged > 65 years (n = 163, 10.75 months) when compared to younger patients (n = 107, 8.22 months) (p = 0.0596, Log-Rank test) (Additional file 2: Figure S2).

In the FAS, median progression free survival (PFS) was 4.37 months. Rash-positive patients achieved a median PFS of 5.06 months, rash-negative patients a median PFS of 4.11 months (p = 0.1412, Log-Rank test) (Fig. 2b). Subgroup analyses revealed a statistically significant benefit in PFS of patients with rash grade ≥ 2 (n = 51, 6.87 months) when compared to patients with rash grade 1 (n = 48, 3.34 months) and patients without rash (n = 171, 4.11 months) (p = 0.0078, Log-Rank test) (Fig. 2c). Furthermore, patients aged > 65 years (n = 163, PFS 5.06 months) benefitted from treatment with gemcitabine/erlotinib when compared to patients ≤65 years of age (n = 107, PFS 3.45 months) (p = 0.012, Log-Rank test) (Fig. 2d).

A multivariate Cox proportional model was used to examine the influence of rash, ECOG performance status at baseline, duration of erlotinib treatment and age on OS and PFS. In the FAS, only age > 65 years proved to be associated with longer OS (hazard ratio 0.640; p = 0.0327) and PFS (hazard ratio 0.642; p = 0.0026) (Additional file 3: Table S1) and this effect was independent of skin rash. Duration of erlotinib treatment was also associated with a favorable outcome with regards to OS and PFS, most probably reflecting a negative selection of patients with early progression leading to short exposure to erlotinib. Response and DCRs did not differ significantly between patients with and without rash (not shown).

Out of the 338 patients in the full safety set, 39 patients (11.5%) completed the pre-specified 12-month observation period, whereas 292 patients (86.4%) terminated the study prematurely (study completion unknown for 7 patients [2.1%]). 133 patients died before regular study end, whereas 130 patients were alive at study termination with data available. Of these 130 patients, 33.1% (n = 43) proceeded to receive further line treatment. Twenty-nine patients were lost to follow-up (8.6%) (Table 3). The type of further line treatment was not documented.

A total of 1681 adverse events (AEs) were recorded in the study. Out of the 338 patients in the SAF, 310 patients (91.7%) experienced at least one AE. A detailed listing of all the grading, outcome and type of AEs is displayed in Additional file 4: Table S2 and Additional file 5: Table S3. Most AEs were CTC grade 1 to 3. In total, 222 patients (65.7%) experienced AEs related to erlotinib, while 156 patients (46.2%) had AEs related to gemcitabine. A special focus was laid on skin-related AEs. In the SAF, 176 patients (52.1%) experienced skin-related side effects, all of which were CTC grade 1 to 3 (Additional file 6: Table S4). Overall, serious adverse events (SAEs) were documented in approximately half of the SAF (50.6%, n = 171) including patients, for whom a tumor progression was documented as SAE (19.2%, n = 65). 68 patients (20.1%) had a fatal SAE including 22 patients (6.5%) for whom progression of the underlying tumor disease was documented as related to the fatal SAE. According to the definition of SAEs in the protocol, a fatal or life-threatening event had to be documented as an SAE.

During treatment, patients self-reported their treatment satisfaction and assessment of side effects at weeks 4, 8 and 16 on a numerical scale from one (very satisfied/less than expected/very good) to six (not satisfied/more than expected/very poor). At week 4, patients were satisfied regarding treatment side effects (mean score: 1.9 ± 0.9, n = 180) and with the information about what to do in case of occurring side effects (mean score: 1.9 ± 0.8, n = 179). With regard to the severity of side effects, patients reported the side effects slightly less severe than what they had expected before the therapy (mean score: 2.6 ± 1.1, n = 179). Patients rated their quality of life as fair to good (mean score: 2.9 ± 1.1, n = 180). Overall, these parameters remained virtually unchanged during the course of the study with exception of the patients’ assessment of the experienced side effects in comparison to anticipated side effects; here patients reported an incremental increase in intensity (Fig. 3).