Prognostic impact of systemic anti-cancer therapy on patients with advanced cancer at end of life: Analysis of a multicenter prospective cohort study using propensity score methods.

Abstract

12029 Background: Prognosis is said to worsen when systemic anti-cancer therapy (SACT) is administered until a patient’s end of life (EOL) and professional palliative care is delayed. However, the prognostic impact of SACT at EOL is unclear. Methods: In this pre-planned secondary analysis of a multicenter, prospective East Asian cross-cultural collaborative cohort study, consecutive patients with advanced cancer admitted to palliative care units between January and December 2017 were divided into four groups: patients who had and had not received SACT (SACT vs. non-SACT) and patients whose last administration of SACT was before or within 1 month of study enrolment (non-1M SACT vs. 1M SACT). The primary endpoint was difference in survival time from a palliative performance scale (PPS) of < 20 to death between the SACT and non-SACT groups, and the 1M SACT and non-1M SACT groups. The analysis was adjusted for age; sex; primary cancer site; metastatic site; comorbidity (Charlson comorbidity index); and history of radiotherapy, smoking, and psychiatric disorders. Weighting for inverse probability of treatment was applied. A secondary endpoint was to identify EOL symptom and care factors prognostic for survival time using a Cox proportional hazards model. Results: The 1396 study patients (712 men, 684 women; median age: 73 years) came from 21 institutions throughout Japan. Primary tumor sites were upper gastrointestinal (471 patients), lower gastrointestinal (193 patients), lung (241 patients), breast (93 patients), urologic (101 patients), gynecologic (82 patients), head and neck (56 patients), central nervous system (25 patients), and hematologic (39 patients). Metastatic sites were liver (554 patients), lung (522 patients), bone (373 patients), and brain (199 patients). The SACT, non-SACT, 1M SACT, and non-1M SACT groups consisted of 853, 539, 126, and 1270 patients, respectively. Survival time was significantly shorter in the SACT group (6.83 days, p = 0.03) than in the non-SACT group (9.01 days). No significant difference between the 1M SACT group (7.57 days) and the non-1M SACT group (7.36 days) was evident. Prognostic factors were use of opioids (HR 1.289), antipsychotics (HR 0.787), and anxiolytics (HR 0.867); infusion of more than 1 L daily (HR 0.727); and apnea (HR 0.874), dyspnea (HR 1.874), nausea (HR 0.612), fatigue (HR 1.284), and ascites (HR 1.236). Conclusions: Prognosis was shorter for patients with a history of SACT than for those without a history of SACT at EOL. However, SACT at EOL was not a prognostic factor. This information is useful in advance care planning for selecting and explaining treatment at EOL to patients and their families.