Risk factors for unplanned discontinuation of scheduled treatment in elderly patients with castration-resistant prostate cancer: results of the IBuTu study

Abstract

Purpose

To gain knowledge about the factors associated with discontinuation of scheduled treatment in elderly men with castration-resistant prostate cancer (CRPC).

Methods

Patients ≥ 70 years with CRPC starting a new line of treatment were included in a prospective cohort study. A geriatric assessment (CGA) was performed at baseline, including comorbidity, mobility, functional/mental/nutritional status, as well as depression. Furthermore, pain intensity, quality of life, ECOG-performance status, and physicians’ and patients’ perception of health were documented. Reasons for and factors associated with discontinuation of scheduled treatment were analysed by univariate and multivariate analysis.

Results

After inclusion of 177 of 300 planned patients, the study was closed due to slow recruitment. 160 patients were eligible for final analysis. Median age was 77.5 years. 46% received chemotherapy, and 54% hormonal treatment. Discontinuation of scheduled treatment occurred in 91 patients (57.6%). The main reasons were progressive disease/death in 63%, adverse events/toxicity in 22%, and withdrawal of consent in 8%. In bivariate analyses, factors associated with discontinuation of treatment were age ≥ 80 years, ECOG PS ≥ 2, compromised/poor health status (physicians’/patients’ assessment), and compromised functional or nutritional status. In multivariate analysis, the only remaining factor independently associated with discontinuation of scheduled treatment was impairment of activities of daily living (ADL < 100 points) (OR = 4.2 for discontinuation; p < 0.05).

Conclusion

Despite limitations due to early termination of the study, our results demonstrate that discontinuation of scheduled treatment was common, and that compromised ADL seems to be a significant risk factor for treatment failure in elderly patients with CRPC.

Introduction

With a median age above 65 years at diagnosis, and two out of three tumour deaths occurring after the age of 75 years, prostate cancer is mainly a disease of the elderly. The International Society of Geriatric Oncology (SIOG) has, therefore, published, and recently updated, recommendations for the management of elderly patients with prostate cancer (Droz et al. 2010, 2014, 2017). The recommendations state that an evaluation of the health status of a patient should include a validated screening tool like the so-called G8 (Bellera et al. 2012; Decoster et al. 2015), and the assessment of comorbid conditions (e.g., using the CISR-G scale), degree of dependence and nutritional status. In frail or comorbid patients, a full geriatric assessment (GA) is recommended.

Elderly patients are underrepresented in clinical trials (Hutchins 1999). Furthermore, there is a lack of data on feasibility and outcome of treatment in elderly patients outside clinical trials. Both under- and overtreatment put patients at risk; therefore, decision making in elderly cancer patients is a challenging task (Pallis 2010). In prostate cancer, both tumour heterogeneity and patient heterogeneity increase the complexity in decision making. This is clinically important, as elderly patients can tolerate treatment generally less well, and show more adverse events than younger patients (for review, see Balducci and Extermann 2000). Comorbidities, decreased functional status (assessed by activities of daily living, ADL, and instrumental activities of daily living, IADL), nutritional status, and cognitive function are all associated with adverse outcomes in patients with prostate cancer. A guideline of the SIOG from 2014 stated that “older men with prostate cancer should be managed according to their individual health status, not according to age” (Droz et al. 2017). These recommendations have recently been updated. Interestingly, the guideline recommends the use of the G8 screening tool not only for patient assessment, but also for making treatment decisions in elderly prostate cancer patients (Droz et al. 2017).

Our study “Interdisziplinäre Behandlung von urologischen Tumoren (IBuTu)” is the initiative of a working group of office-based urologists and oncologists in Germany, exploring clinical aspects of the interdisciplinary management of prostate cancer patients. To gain more insight into factors associated with treatment outcome, the group realized a prospective registry for elderly prostate cancer patients with castration-resistant prostate cancer (CRPC). Outside clinical trials, cohort studies and registries are valuable tools for the collection of data on characteristics and outcomes of patients with malignancies (Wildiers et al. 2013). The main aim of our cohort study was to identify frequency and factors associated with unplanned discontinuation of systemic treatment (feasibility of treatment). Clinically, the probability of a good or poor feasibility of treatment is of major importance; however, feasibility of treatment is so far not a very well-defined endpoint in medical oncology (Wildiers et al. 2013).

Objective and design

The research hypothesis was that a GA, or elements thereof, is superior to chronological age and conventional clinical judgement in predicting whether or not a patient with CRPC is likely to receive a full course of scheduled systemic tumour therapy (antihormonal treatment, AHT, or chemotherapy, CTX). We carried out a prospective multicentre, non-interventional cohort study to investigate the prognostic validity of a GA in elderly patients with CRPC. Endpoint was any unplanned treatment discontinuation of scheduled systemic therapy. Reasons for discontinuation were disease progression, death, severe adverse events, toxicity, and patient’s wish to stop treatment.

Patients and methods

Data was collected from patients ≥ 70 years starting a new line of treatment, either chemotherapy, newer androgen receptor targeting agents, or alpharadin (Table 1, Patient characteristics). Any line of treatment was allowed. Besides clinical and demographical data, a GA was performed, including comorbidity assessment (Charlson Score), balance and mobility tests (Timed Up and Go, and Tinetti), ADL, IADL, mini-mental state examination (MMSE), depression screening (PHQ-9), and the mini nutritional assessment (MNA). Pain was assessed using the visual analogue scale (VAS), and quality of life and patient self-assessed state of health was measured using the SF12-test. Furthermore, ECOG-performance status and physicians’ rating of overall health status of the patient (very good, good, fair, compromised) were documented; in addition, PSA-kinetics and preceding treatment were included. Reflecting clinical practice of treatment duration, which usually is limited to 8–10 cycles of chemotherapy (i.e., 24–30 weeks) and unlimited duration of antihormonal treatment, patients receiving chemotherapy (CTX) were followed for 6 months, and patients receiving newer antihormonal treatment (AHT) were followed for 12 months (end of observation).

Table 1.

Patient characteristics

Variable	Total (n)	Total (%)
Age group
70–74	43	26.88
75–79	73	45.63
≥80	44	27.50
Total	160	100.00
Therapy group
12 months	87	54.38
6 months	73	45.63
Total	160	100.00
Chosen medicationa
Abiraterone acetate	69	43.95
Cabazitaxel	20	12.74
Docetaxel	50	31.85
Enzalutamid	18	11.46
Total	157	100.00
Charlson score
0–5 Pt	50	31.65
6–8 Pt	75	47.47
≥ 9 Pt	33	20.89
Total	158	100.00
ECOG
≥ 2 Pt	31	19.50
1 Pt	87	54.72
0 Pt	41	25.79
Total	159	100.00
Health status
Restricted	16	10.00
Average	33	20.63
Good	79	49.38
Very good	32	20.00
Total	160	90.00

^aOnly for the four most common medications, there are three additional medications used by only one patient

Statistical analysis

The study was designed to test the following hypotheses: (1) GA is superior to chronological age or physicians’ assessment in predicting unplanned treatment termination (failure of treatment). (2) Suitable tests for assessing the risk of failure are the Tinetti test, Timed Up and Go, nutritional assessment (weight loss), the PHQ-9 test (risk of depression), Charlson score (comorbidities), and the SF-12 test (patients’ perception of health). To achieve sufficient statistical power, 300 patients had to be included. Descriptive frequency tables and descriptive statistics regarding patients’ characteristics, health status, course of therapy, outcome of treatment (therapy discontinuation, successful treatment) were performed. All independent variables underwent bivariate screening (p < 0.2) before being included in a logistic regression model. In addition, the backward selection procedure was used to create a final logistic regression model with significant variables (p < 0.05). Study endpoint was any unplanned treatment discontinuation of systemic tumour therapy due to progressive disease (defined as pain due to tumour, increasing PSA, or progression on imaging, adverse events, toxicity, or patients’ withdrawal of consent within 6 months for patients with CTX, and within 12 months for patients with AHT. The study was reviewed and approved by the institutional review board of Charité University Hospital Berlin, Germany).

Results

Patient characteristics and treatment

Between 07/2013 and 12/2015, 177 of 300 planned patients were included after obtaining informed consent into the study from 38 centres, 7 oncologic and 31 urologic practices. Due to slow accrual, the study had to be terminated early. Data of 17 patients had to be excluded for various reasons (age below age cut-off; death before trial entry; GA not done; application of non-standard therapy or no therapy). 160 patients were eligible for analysis, with missings in some of the variables, and for 98 patients, a final assessment was available at end of study, see Fig. 1, consort diagram.

Fig. 1.

Flow diagram of the progress through the phases of the IBuTu study

46% received chemotherapy (either docetaxel or cabazitaxel), and 54% received antihormonal treatment with abiraterone acetate or enzalutamide. One patient each received estramustin or alpharadin. 80% of patients had an ECOG PS 0–1, 20% an ECOG PS ≥ 2. Median age was 77.5 years. Patient characteristics are summarized in Table 1. The distribution of results of the CGA is summarized in Table 2.

Table 2.

Results of the comprehensive geriatric assessment (CGA)

Variable	Total (n)	Total (%)
ADL
< 100 Pt	60	40.82
100 Pt	87	59.18
Total	147	100.00
IADL
< 8 Pt	47	33.10
8 Pt	95	66.90
Total	142	100.00
MMSE
< 24 Pt	12	8.33
24–30 Pt	132	91.67
Total	144	100.00
Tinetti
< 28 Pt	88	63.77
28 Pt	50	36.23
Total	138	100.00
TUG
> 10 s	99	70.21
≤ 10 s	42	29.79
Total	141	100.00
MNASF
< 14 Pt	90	61.64
14 Pt	56	38.36
Total	146	100.00
PHQ_9_1
Healthy/unremarkable	111	77.08
Depressive	33	22.92
Total	144	100.00
VAS
No pain	56	38.10
Slight pain	77	52.38
Severe pain	14	9.52
Total	147	100.00
Balducci
Group 1 and 2	26	18.57
Group 3	114	81.43
Total	140	100.00
BMI
Normal weight (< 25)	45	30.61
Overweight (25–30)	66	44.90
Adiposity (> 30)	36	24.49
Total	147	100.00
PCS_1
1. Quartile (< 31.57)	33	24.09
2. Quartile (31.57–< 38.57)	35	25.55
3. Quartile (38.37–< 50.22)	35	25.55
4. Quartile (≥ 50.22)	34	24.81
Total	137	100.00
MCS_1
1. Quartile (< 38.55)	33	24.09
2. Quartile (38.55–< 48.79)	35	25.55
3. Quartile (48.79–< 57.74)	35	25.55
4. Quartile (≥ 57.74)	34	24.81
Total	137	100.00

Unplanned discontinuation of scheduled treatment

Unplanned discontinuation of treatment was reported in 91 of 160 patients (56.8%). 57% of patients receiving antihormonal treatment discontinued treatment within 12 months, and 43% of patients receiving chemotherapy discontinued treatment within 6 months. Main reasons were progression/death in 63%, adverse events/toxicity in 22%, and patients’ wish 8% (Table 3).

Table 3.

Reasons for unplanned discontinuation

Reasons for unplanned treatment discontinuation of therapy (grouped)	Freq.	Percent
Tumour progression	48	52.75
Other AE or SAE	19	20.88
Death of patient	9	9.89
Patients decision	7	7.69
Lost to followx-up	5	5.49
Toxicity	1	1.10
Secondary disease	1	1.10
Not stated	1	1.10
Total	91	100.00

In bivariate analyses, factors associated with treatment discontinuation were age ≥ 80 years, ECOG PS ≥ 2, compromised/poor health status (physicians’/patients’ assessment), and compromised functional or nutritional status, see Table 4.

Table 4.

Factors associated with treatment discontinuation

Variable	Unplanned treatment discontinuation of therapy			p valueb
	Yes		Total
Age group
70–74	22	51.16%	43	0.045
75–79	37	50.68%	73
≥ 80	32	72.73%	44
Total	91	56.88%	160
Therapy group
12 Months	52	59.77%	87	0.420
6 Months	39	53.42%	73
Total	91	56.88%	160
Chosen medicationa
Abiraterone acetate	40	57.97%	69	0.079
Cabazitaxel	15	75.00%	20
Docetaxel	22	44.00%	50
Enzalutamid	12	66.67%	18
Total	89	56.69%	157
Charlson Score
0–5 Pt	28	56.00%	50	0.609
6–8 Pt	40	53.33%	75
≥ 9 Pt	21	63.64%	33
Total	89	56.33%	158
ECOG
≥ 2 Pt	23	74.19%	31	0.037
1 Pt	49	56.32%	87
0 Pt	18	43.90%	41
Total	90	56.60%	159
Health status
Compromised/poor	13	81.25%	16	0.018
Average	22	66.67%	33
Good	44	55.70%	79
Very good	12	37.50%	32
Total	91	56.88%	160
ADL
< 100 Pt	44	73.33%	60	0.001
100 Pt	40	45.45%	88
Total	84	56.76%	148
IADL
< 8 Pt	32	68.09%	47	0.053
8 Pt	49	51.04%	96
Total	81	56.64%	143
MMSE
< 24 Pt	9	75.00%	12	0.178
24–30 Pt	73	54.89%	133
Total	82	56.55%	145
Tinetti
< 28 Pt	55	61.80%	89	0.072
28 Pt	23	46.00%	50
Total	78	56.12%	139
TUG
> 10 s	56	56.57%	99	0.735
≤ 10 s	23	53.49%	43
Total	79	55.63%	142
MNASF
< 14 Pt	57	62.64%	91	0.054
14 Pt	26	46.43%	56
Total	83	56.46%	147
PHQ_9_1
Healthy/unremarkable	59	53.15%	111	0.136
Depressive	23	67.65%	34
Total	82	56.55%	145
VAS
No pain	29	51.79%	56	0.593
Slight pain	46	58.97%	78
Severe pain	9	64.29%	14
Total	84	56.76%	148
Balducci
Group 1 and 2	13	48.15%	27	0.316
Group 3	67	58.77%	114
Total	80	56.74%	141
BMI
Normal weight (< 25)	29	63.04%	46	0.106
Overweight (25–30)	40	60.61%	66
Adiposity (> 30)	15	41.67%	36
Total	84	56.76%	148
PCS_1
1. Quartile (< 31.57)	26	76.47%	34	0.026
2. Quartile (31.57–< 38.37)	22	62.86%	35
3. Quartile (38.37–< 50.22)	19	54.29%	35
4. Quartile (≥ 50,22)	14	41.18%	34
Total	81	58.70%	138
MCS_1
1. Quartile (< 38.55)	21	61.76%	34	0.789
2. Quartile (38.55–< 48.79)	18	51.43%	35
3. Quartile (48.79–< 57.74)	21	60.00%	35
4. Quartile (≥ 57.74)	21	61.76%	34
Total	81	58.70%	138

^aOnly for the four most common medications, there are three additional medications used by only one patient

^bp value of Chi-squared test

In multivariate analysis (backward selection), the only remaining factor independently associated with treatment discontinuation was compromised functional status. Patients achieving the full score of 100 points in ADL score showed an odds ratio (OR) of 0.24 for discontinuation (confidence interval 0.104–0.554, p < 0.05) compared to those who did not achieve the full score of 100 points.

We performed an exploratory analysis excluding all patients with progression of disease. In the remaining cohort of 43 patients, ADL lost its significance as a risk factor of treatment failure; on the other hand, age cohort showed significant association with treatment failure. Patients aged 70–74 years had a 72% lower risk of unplanned treatment discontinuation than patients aged ≥ 80 years (HR 0.28; confidence interval 0.096–0.809, p < 0.05).

Discussion

To gain more insight into factors associated with unplanned discontinuation of scheduled systemic tumour therapy in elderly patients with CRPC treated outside tertiary care or academic centres, we performed the IBuTu study, a prospective data collection of patients treated in community-based urologic and oncologic practices in Germany. Since the recruitment goal of 300 patients was not achieved within the trial period of 24 months, the trial had to be terminated early. However, with 160 patients included, an exploratory analysis using bivariate and multivariate analyses was performed to assess factors associated with unplanned treatment discontinuation.

Besides characteristics routinely assessed in oncologic patients such as ECOG, age cohort, or choice of treatment, we also assessed factors of a GA, as well as patients’ and physicians’ assessment of health. As documented by patient characteristics in Table 1, our patients represent a typical population of outpatients with CRPC above 70 years, in community practices.

Several studies have assessed factors associated with treatment tolerance and all-cause mortality in older patients with cancer, and identified, among others, age, poor nutritional status, impaired mobility, and impaired mini mental status as risk factors (Extermann et al. 2012). A systematic review recently analysed available data regarding the value of GA for predicting patients’ outcome, and concluded that some variables were of value, though the results were somewhat inconsistent (Hamaker et al. 2012). A randomized phase III trial of patients with advanced lung cancer using the GA classification (fit, vulnerable, and frail) to guide treatment decisions, failed to show an improvement in time to treatment failure, or overall survival, compared to patients in the standard arm; however, a decrease of all grade toxicity and fewer treatment failures due to toxicity were observed (Corre et al. 2016).

So far, factors associated with, or predictive of, unplanned treatment discontinuation in patients with CRPC cancer have not been assessed extensively. Della Pepa et al. recently reported results of a small series of 24 patients ≥ 70 years receiving docetaxel chemotherapy (Della Pepa et al. 2017). While they found a statistically significant relationship between frailty (as assessed by CGA) and early docetaxel discontinuation, association between frailty and response to chemotherapy was non-significant.

In our analysis, the factors that emerged from bivariate analysis as being associated with treatment discontinuation, were age ≥ 80 years, ECOG PS ≥ 2, compromised/poor health status (physicians’/patients’ assessment), and compromised functional or nutritional status. Interestingly, the only factor emerging from a logistic regression analysis as independently associated with treatment discontinuation was compromised functional status, i.e., less than the full score of 100 points in the activities of daily living (ADL). In patients with a compromised ADL score, the risk of discontinuation was more than fourfold higher compared to those with a full ADL score.

Excluding cases with unplanned treatment discontinuation due to progression of disease yielded a different picture: while ADL did not reach significance as an independent factor, age was the only factor significantly associated with treatment failure. However, due to small patient numbers, our data must be interpreted with caution, and more analyses with sufficient patients to assess factors associated with treatment outcome are warranted in CRPC.

Our results support the notion that a GA, or elements thereof, can provide the physician and the patient with additional information that can be valuable for decision making in geriatric oncology. In particular, assessing ADL might be of value in patients with CRPC before starting systemic treatment. As current guidelines recommend the use of the geriatric screening tool G8, which does not incorporate ADL assessment, it would be interesting to compare ADL and G8 in a prospective manner in elderly CRPC patients at the start of a new treatment.

Our study has several limitations. A relevant restriction is the fact that the study had to be terminated early due to slow recruitment. This limits the statistical power of the study and left us with a descriptive analysis. Furthermore, our decision to include patients whose treatment was stopped due to progressive disease into the group with early treatment discontinuation can be criticised. It would of course be interesting to separately analyse factors associated with unplanned treatment discontinuation (both including and excluding tumour progression) in future studies with sufficient statistical power.

In summary, we conclude that recruitment of patients into this non-interventional cohort study from community practices in Germany was more challenging than anticipated. We postulate, however, that despite statistical limitations due to the early termination of the study, compromised ADL could be an easily accessible risk factor identifying elderly patients with CRPC at risk of unplanned discontinuation of scheduled treatment.

Funding

This study was funded by Janssen-Cilag GmbH (Grant number 212082PCR4009).

Compliance with ethical standards

Conflict of interest

Author Friedemann Honecker declares that he has no conflict of interest. Author Ulrich Wedding declares that he has no conflict of interest. Gerd Kallischnigg declares that he has no conflict of interest. Author Axel Schroeder declares that he has no conflict of interest. Author Jörg Klier declares that he has no conflict of interest. Author Thomas Frangenheim declares that he has no conflict of interest. Author Lothar Weißbach declares that he has no conflict of interest.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This article does not contain any studies with animals performed by any of the authors.

Informed consent

Informed consent was obtained from all the individual participants included in the study.