Patient preferences for palliative treatment of locally advanced or metastatic gastric cancer and adenocarcinoma of the gastroesophageal junction: a choice-based conjoint analysis study from Germany

To our knowledge, this study provides the first patient preference data for a new hypothetical palliative CT of gastric cancer, performed after patients had started treatment. All patients were able to complete the CBC module, and most (92.7%) perceived the complete survey as a positive or very positive experience, confirming that CBC analysis can be appropriately used in these severely ill patients. In this CBC analysis, treatment tolerability and the ability to self-care were ranked highest in importance by a sample of 55 patients with mGC or mGEJ-Ca and varied CT experience over the last 2 years. A palliative CT associated with no or mild adverse reactions and requiring no hospitalization was considered twice as important as an additional 3-month survival benefit, and requiring little or no assistance for daily living activities was considered 1.5 times as important as an additional 3-month survival benefit. The findings indicate that patients with previous CT experience consider a survival benefit accompanied by high quality of life, i.e. being able to self-care and receiving a treatment with good tolerability, as more important than an additional survival benefit per se. In direct questioning, the importance of survival was perceived higher than in the CBC analysis, yet the weighted responses of patients trading off between different aspects of their daily life, disease and treatment in the CBC model provide a broader picture and should therefore be considered as more complete when evaluating patient preferences. In the end, this interpretation is consistent with the results of the 6 qualitative interviews, and with the results from direct, open-ended questioning, where goals related to prolonged survival (prolonged survival, cure, or gaining time) were most frequently mentioned as the most important treatment goals, followed by avoiding disease progression or achieving tumor shrinkage, and treatment goals related to symptom improvement (improved overall performance, no limitations in daily routine, pain-free living). Nevertheless, physicians should be aware that they need to word their questions carefully when trying to identify their patients’ true preferences. Patient preferences may have differed depending on patients’ main treatment goals. However, the sample size (N?=?55) precluded any subgroup analysis by treatment goal.

Patient preferences have been previously evaluated for other tumor entities such as breast cancer or non-small cell lung cancer (NSCLC) [23, 24]. These studies indicate that preferences may differ considerably, depending on factors such as tumor type, severity of disease, and extent of previous treatment. For example, a CBC study in 121 patients with Stage I-IV breast cancer, all treated with CT during the last 5 years, identified a survival benefit of 3 months as the most important preference. These patients considered a more convenient administration regimen as less important than a 13% chance or more of severe toxicities, but more important than a 10–12% chance of severe toxicities [23]. In another recent study, 211 patients with NSCLC who had been treated within the last 2 years considered an increase in progression-free survival as the most important factor, followed by a reduction in tumor-associated symptoms (cough, shortness of breath, and pain), and the reduction of side effects. Mode of administration was considered as least important.

Subgroup analyses revealed that the relative importance of “progression-free survival” increased with therapy experience [24].

In all these previous investigations, as well as the current study, patients were already exposed to CT before patient preferences were assessed. In this study, more than 80% of patients were currently receiving CT when they completed the survey. This limits the informative value for the strategic decision for or against CT based on median survival data from randomized clinical trials as patients still do not have any experience of the potential benefits and toxicities. Yet, the results give hints for patients’ preferences when choosing between different treatment options. Also, patients who have previously experienced a palliative benefit (e.g. improved dysphagia) or tumor response can be expected to be more in favor of CT than patients who had progressive disease and experienced adverse reactions. In addition, untreated, less severely ill patients might consider adverse reactions as less important and survival benefit as more important than patients currently suffering from adverse reactions during CT. Because performance status, the treatment regimen given, the timing of the survey in relation to patients’ ongoing CT (i.e. during recovery period between cycles or during acute toxicity phase), tumor response, and toxicity data were not captured in this survey, their impact on patient preferences could not be assessed. This might be considered as considerable limitation. On the other hand, including different patients with different CT experiences may help to mirror the real-life situation more closely. Further, the study included only patients who were willing to participate and were considered fit enough for participation by their physician. Therefore, the study population may not be representative of the general population of gastric cancer patients receiving palliative CT.

Another limitation of the study is that while the maximum additional survival benefit over standard of care in the fictive patient profiles of the conjoint analysis survey was 3 months which reflects the differences between various modern CT regimens (older vs. more modern, doublet vs. triplet), modern first-line CT regimens offer a more pronounced survival benefit of up to 9 months over best supportive care [7–10]. Finally, the sample size was limited in our study, and it cannot be excluded that the recruitment procedure (treating physicians contacted target patients) may have resulted in selection bias. On the other hand, the high root likelihood values for both models (MLR and HB) and the consistency of results across different model approaches indicate that the CBC analysis provided high-validity results.