Patients were enrolled according to the following inclusion criteria: (1) age 65 years or older, (2) newly diagnosed, pathologically proven solid cancer, and (3) life expectancy of ≥ 3 months. Patients with psychiatric disorders or serious uncontrolled comorbidities or who could not communicate were excluded. Written informed consent was obtained from each patient before enrollment. We enrolled patients being treated at Uijeongbu St. Mary's Hospital between April 2012 and November 2013. The questionnaires for geriatric assessment were administered by one professional nurse before the initiation of the planned treatment. The clinical and laboratory data at the time of diagnosis before treatment were collected. The following laboratory tests were performed and the values used to evaluate the relationships with the CGA: complete blood count, serum levels of albumin, creatinine, lactate dehydrogenase (LDH), C-reactive protein (CRP), and D-dimer. The patients received treatment and were followed up until death or December 2015. Treatment interruption group included patients who had discontinued treatment during their first-line therapy or who had not received second-line chemotherapy after progression for the following reasons: (1) physicians determined that additional treatment was not appropriate due to deterioration in general condition of the patient and (2) patients refused the continuous treatment due to treatment-related toxicity. Ethical committee approval was obtained from the Institutional Review Board of The Catholic University of Korea, Uijeongbu St. Mary’s Hospital (UC12QISI0013).

The geriatric assessment included the following instruments: Korean version of the Modified Barthel Index (K-MBI), Korean version of Instrumental Activities of Daily Living (K-IADL), Mini-Mental State Examination (K-MMSE), Geriatric Depression Scale (K-GDS), and the European Organisation for Research and Treatment of Cancer Quality-of-Life Core Questionnaire (EORTC-QLQ-C30). All questionnaires were previously validated in the Korean language [5-8]. First, we used the K-MBI (SHAH version) to assess performance in activities of daily living. This comprises 10 items and yields a score of 0 to 100; a score of 100 is interpreted to indicate independence in activities of daily living [5]. Second, we used the K-IADL to evaluate the instrumental functions necessary for adaptation to social life. Patients with at least one dependency in IADL were classified as IADL dependent. Third, we used the K-MMSE as a screening tool for cognitive dysfunction in the elderly patients. The scores range of 0 to 30. For the present study, we adopted a single cutoff and classified the range of 0 to 24 as definitive cognitive dysfunction [6,9]. Fourth, mood was assessed using the K-GDS, which comprises 30 questions and is scored range from 0 to 30. Scores ≥ 16 were classified as indicating a depressive disorder [7].

We used the EORTC-QLQ-C30 to assess the quality of life. The EORTC-QLQ-C30 (version 3.0) is a quality-oflife questionnaire used extensively in patients with cancer. It comprises 30 questions about global health status, functioning scales (physical, role, emotional, cognitive, and social), and symptom scales (fatigue, nausea/vomiting, pain, dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties). Raw scores for each scale were linearly transformed into a score of 0 to 100. A high score for the global health status and functioning scales represents a better level of functioning and a high quality of life, whereas a high score for symptom scales represents worse symptoms [8].

Continuous variables were expressed as mean with standard deviation. Continuous variables were compared using the t test or nonparametric Mann-Whitney U test. The chi-square test and Fisher exact test were used to compare categorical variables. Univariate and multivariate logistic regression analyses were performed to compare between the treatment interruption and continuation groups. To determine the significance of bivariate associations, we used Pearson correlation coefficient. Progression free survival (PFS) was defined as the time from start date of chemotherapy to disease progression or death from any cause. Overall survival (OS) was defined as start date of chemotherapy until death from any cause. Univariate survival analysis was performed using the Kaplan-Meier method, and the log-rank test was applied to identify significant differences. Multivariate survival analysis was performed using the Cox proportional-hazard regression model. Statistical analyses were performed using IBM SPSS version 19.0 (IBM Co., Armonk, NY, USA) and p < 0.05 was considered significant.

We enrolled 111 elderly cancer patients. However, 20 patients with stage II or III cancer, nine patients who did not receive treatment, and seven patients who received targeted or hormone therapy were excluded from the final analysis. Therefore, 75 patients were included in this study. The general characteristics of the 75 patients are summarized in Table 1. The mean age of the patients was 72.7 years (standard deviation, 5.0). Fifty-three patients (70.6%) had one or more comorbid disease, and the most common comorbidity was hypertension (60.0%). The numbers of patients with diabetes, stroke, cardiovascular disease, and pulmonary disease were 15 (20.0%), five (6.7%), five (6.7%), and three (4.0%), respectively. The most frequent cancer types were gastrointestinal (28.0%) and lung/head and neck (21.3%) cancers. Twenty-five (33.1%) had metastases to two or more organs, and the most common sites of distant metastases were liver (30.7%), lung (28.0%), distant lymph nodes (28.0%), and bone (26.7%).

Twenty patients (26.6%) received monochemotherapy and 55 patients (73.4%) received combination-chemotherapy. The median number of cycles of chemotherapy was 3 (range, 1 to 12). Forty patients (53.3%) received fulldose chemotherapy and 19 patients (25.3%) received a reduced starting dose of chemotherapy.

Among the 75 patients, 41 patients (54.6%) stopped cancer treatment during the first-line therapy or after end of first-line therapy because of death, deteriorated condition, or the patient’s refusal and these patients were classified as the treatment interruption group. Twenty patients stopped treatment due to death, and 10 and 11 patients discontinued therapy due to deteriorated condition and patient’s refusal, respectively. The median PFS was 3.7 months (range, 0.17 to 11.4) in the treatment interruption group and 4.8 months (range, 0.77 to 34.2) in the continuation group.

The geriatric assessment and quality of life scores of patients are summarized in Table 2. The K-MBI, K-IADL, and EORTC-QLQ-C30 were administered to all patients, and the K-MMSE and K-GDS were administered to 74 patients. Thirty-seven (49.3%) and 33 patients (44.0%) were classified as dependent in activities of daily living and instrumental activities of daily living at the time of diagnosis, respectively. We estimated that 45 patients (60.8%) had some degree of decline in cognitive function, and 19 patients (25.7%) had a depressive disorder.

We evaluated the relationships between the geriatric assessment components and clinical and laboratory parameters. The K-MBI correlated negatively with age and the K-GDS, and correlated positively with the EORTC-QLQ-C30 global health status and the physical and role functioning scales. Patients with a cognitive dysfunction identified by the K-MMSE had a lower educational level. K-GDS correlated negatively with the EORTC-QLQ-C30 scores for global health status, physical, role, emotional, and cognitive functioning scales. The scores for the physical functioning scales correlated positively with the level of albumin but negatively with the levels of LDH, CRP, and D-dimer (Supplementary Table 1).

The results of the univariate and multivariate analyses to compare patients in the treatment interruption and continuation groups are listed in Table 3. The clinical and laboratory variables that were significantly associated with treatment interruption were ≥ 2 sites of distant metastases and lower level of albumin. Patients with visceral organ metastases such as in the liver, lung, or peritoneum were more likely to be in the treatment interruption group compared with patients with non-visceral organ metastases. In the geriatric assessment, there were no differences in K-MBI, K-MMSE, and K-GDS between the continuation and interruption group. More patients in the treatment interruption group showed a dependent activity of K-IADL, but there was no statistically significant difference. The patients with treatment interruption showed the significantly lower scores for the EORTC-QLQ-C30 physical and role functioning scales and higher score for the EORTC-QLQ-C30 fatigue and appetite loss symptom scale compared to patients with treatment continuation.

In the multivariate analysis, the variable that was significantly associated with treatment interruption was a low score for the EORTC-QLQ-C30 physical functioning scale (odds ratio [OR], 1.020; 95% confidence interval [CI], 1.002 to 1.039; p = 0.030), and ≥ 2 sites of distant metastases (OR, 2.965; 95% CI, 1.012 to 8.681; p = 0.047)

Sixty-three of the 75 patients (84.0%) died during the follow-up period, and the median OS of all patients was 7.6 months. Forty (97.6%) and 23 patients (67.6%) died in the treatment interruption and continuation groups, respectively. The median OS were 5.7 and 11.4 months in the treatment interruption and continuation groups, respectively. The significant factors related to OS identified in the univariate analysis are summarized in Table 4. In the multivariate analysis, patients with distant metastases in two or more organs, treatment interruption, lower scores for the EORTC-QLQ-C30 physical functioning scales, and high level of D-dimer had a shorter OS (Table 5, Fig. 1).