JOURNAL OF PALLIATIVE MEDICINE

Volume 17, Number 3, 2014

ª Mary Ann Liebert, Inc.
DOI: 10.1089/jpm.2013.0366

Is There Any Association between Where Patients

Spend the End of Life and Survival after Anticancer
Treatment for Gynecologic Malignancy?

Hiroaki Kajiyama, MD, PhD,1 Fumi Utsumi, MD,1 Makiko Higashi, MD, PhD,2 Jun Sakata, MD,1
Ryuichiro Sekiya, MD, PhD,1 Mika Mizuno, MD, PhD,1 Tomokazu Umezu, MD, PhD,1
Shiro Suzuki, MD, PhD,1 Eiko Yamamoto, MD, PhD,1 Hiroko Mitsui, MD, PhD,1
Kaoru Niimi, MD, PhD,1 Kiyosumi Shibata, MD, PhD,1 and Fumitaka Kikkawa, MD, PhD1

Abstract
Background: It remains unknown whether the end-of-life (EOL) environment influences survival after anti-
cancer treatment, particularly for gynecologic malignancy.
Objective: The study’s objective was to clarify whether the survival time varied depending on where patients
spend the EOL.
Methods: This retrospective study included patients who received initial oncologic treatment but died due to
cancer recurrence and/or progression. The subjects were a cohort of 181 gynecologic malignant tumor cases in a
single institution from 2002 to 2008. Measurement was of postcancer treatment survival (PCS), defined as the
time interval between the last date of anticancer treatment after recurrence/progression and death from the
disease, analyzed on stratification by type of supportive care or where patients spent the EOL.
Results: The median survival time was 26.1 (1.0–306.4) months. The distribution of the carcinoma type was as
follows: 28.7% of patients with cervical (N 52), 27.6% with endometrial (N 50), and 43.1% with ovarian
(N 79) cancer. The median PCS was 13.3 weeks. Patients in the hospice/home care group showed a signifi-
cantly more favorable PCS than those in the hospital group (log rank: P 0.029). On multivariate analysis, the
age ( < 60 versus ‡ 60) and site of supportive care (hospital versus hospice/home care) retained their signifi-
cance as independent prognostic factors of poor PCS (age: HR 0.679, 95% CI, 0.496–0.928, P 0.0151; site of
supportive care: HR 0.704, 95% CI, 0.511–0.970, P 0.0319).
Conclusions: Our current data could be hypothesis generating; it is possible that the EOL environment is a
crucial prognostic factor for survival after anticancer treatment.

Introduction numerous side effects and an increased rate of hospitaliza-

tion. Recurrence sites of gynecologic tumors are diverse in

G ynecologic malignancies include various types of

tumor with diverse oncologic outcomes. Once patients
with these tumors experience recurrence, cure is difficult.
However, at the time of disease recurrence there are not only
salvage strategic anticancer treatments, but also multiple
options for palliative therapies for relapsed patients. In other
words, although the majority of relapsed patients continu-
ously or intermittently receive aggressive therapies, includ-
ing chemotherapy and/or radiotherapy, the end of life (EOL)
in patients with relapsed gynecologic cancer may be fre-
quently impaired by such therapeutic applications due to
systemic organs, including the brain, lungs, liver, bone,
lymph nodes, peritoneum, and pelvis. Since patients suffer
tumor-related symptoms specific to those relapsed regions,
palliative care is frequently accompanied by aggressive
treatment. However, it is one of the difficult issues knowing
when to withdraw aggressive therapy and switch to best
supportive care to improve EOL quality. However, does an
earlier use of opioids, sedatives, and hospice enrollment lead
to a shorter survival time of patients than aggressive treat-
ment? Abundant evidence demonstrated that, with symptom
relief using opioids, a higher dose was not detrimental but

1
Department of Obstetrics and Gynecology, Nagoya University Graduate School of Medicine, Nagoya, Japan.
2
Department of Obstetrics and Gynecology, Tsushima Municipal Hospital, Aichi, Japan.
Accepted October 22, 2013.

325
326 KAJIYAMA ET AL.

actually beneficial to improve the subsequent oncologic hospital, hospice, and home care groups. The university
outcome.1–5 Additionally, based on larger-scale retrospective hospital cohort included patients who continuously received
analysis, for certain terminally ill patients, hospice enroll- anticancer therapy and subsequent palliative care from initial
ment may be associated with longer survival.6 Furthermore, treatment until death in Nagoya University Hospital. Each
Temel and colleagues randomly assigned patients with newly general hospital, hospice, or home care cohort included those
diagnosed metastatic nonsmall cell lung cancer to receive who received primary therapy and thereafter were referred/
either early palliative care integrated with standard oncologic moved to a neighboring general hospital, hospice, or home
care or standard oncologic care alone. Although fewer pa- care at the cessation of the anticancer treatment, respectively.
tients in the early palliative care group than in the standard The survival curves were based on the Kaplan Meier method
care group received aggressive EOL care, the median sur- and compared employing the log rank test. Furthermore, we
vival was longer among patients receiving early palliative analyzed the PCS and ICS on stratification to two subgroups:
care (11.6 versus 8.9 months, respectively, P = 0.02).7 the hospital group (university/general hospital) and hospice/
Nevertheless, it remains unknown whether the EOL envi- home care group. Multivariable analysis was carried out with
ronment influences survival after anticancer treatment, par- the Cox proportional hazards model to evaluate independent
ticularly for gynecologic malignancy, which includes the factors affecting survival. A P-value of < 0.05 was consid-
above-mentioned symptoms. Thus, in this study, we analyzed ered significant.
a cohort of 181 cases with gynecologic malignant tumors,
such as cervical, endometrial, and ovarian cancer, from a Results
single institution. The purpose of the current study was to
Patients’ characteristics
clarify whether the survival time varied in response to where
patients spent the EOL. The clinical characteristics of the 181 patients enrolled are
described in Table 1. The median survival time was 26.1
months, range 1.0 to 306.4 months. The median age at the end
Methods
of aggressive cancer treatment was 58 years old (21 84
Patients years). When we classified patients into the two subgroups on
stratification by the age near the median value, < 60 years in
A retrospective review of all gynecologic cancer patient
98 (54.1%) patients and ‡ 60 years in 83 (45.9%) patients.
deaths from 2002 to 2008 at Nagoya University Hospital was
The distribution of carcinoma types was as follows: 52
performed. The study population included patients who re-
(28.7%) with cervical, 50 (27.6%) with endometrial, and 79
ceived initial oncologic treatment at our institution but died
(43.1%) with ovarian carcinomas. Pelvis invasion was posi-
due to cancer recurrence and/or progression. This study was
tive in 68 (37.6%) and negative in 113 (62.4%) patients. In
approved by the ethics committee in Nagoya University.
152 (84.0%) patients, an opioid was used for pain control. As
Patients were excluded from this study if they had insufficient
the converted oral morphine intake, the medium value of
clinical data, a history of other malignancies, no information
maximum opioid use was 24 mg. Thus, when we divided
on cancer recurrence or death not attributed to gynecologic
patients into the two subgroups on stratification by the
cancer, or were lost to follow-up immediately after the initial
maximum opioid use, the distributions were as follows:
treatment. A variety of clinical records were collected, in-
£ 25 mg in 63 (34.8%) patients and > 25 mg in 89 (49.2%)
cluding demographic data, cancer diagnosis, treatment type,
further treatments if there was recurrence or progression at
diagnosis, the type of supportive care, where patients spent
the EOL, and the date of death. Pelvis invasion was defined as Table 1. Patients’ Characteristics
the presence of a deeply invaded tumor in the pelvic organs
and/or retroperitoneal cavity, excluding superficial peritoneal Total %
dissemination. Maximum opioid use was calculated as the Total 181
oral morphine intake. The calculation formula of each Age
equianalgesic dose to oral morphine based on 24-hour opioid < 60 98 54.1
requirements was as follows: parenteral morphine (1:2), ‡ 60 83 45.9
modified-release oxycodone (20:30), transdermal fentanyl Carcinoma
(1:30), parenteral fentanyl (0.6:0.60), morphine suppository Cervical 52 28.7
(40:60). Endometrial 50 27.6
Ovarian 79 43.6
Survival analysis Pelvis invasion
Positive 68 37.6
Postcancer treatment survival (PCS) was defined as the Negative 113 62.4
time interval between the last date of anticancer treatment
and death from the disease. Furthermore, intracancer treat- Maximum opioid usea
None 29 16.0
ment survival (ICS) was defined as the time interval between £ 25 mg 63 34.8
the date of initial treatment and the last date of anticancer > 25 mg 89 49.2
treatment. In this context, ‘‘cancer treatment’’ included
End-of-life site
initial/cytoreductive/salvage surgery, any chemotherapy, University/general hospital 118 65.2
and/or aggressive radiotherapy, but excluded palliative irra- Hospice/home care 63 34.8
diation. Regarding where patients spent the EOL, we divided
a
all patients into four cohorts: university hospital, general All intakes were converted to oral morphine.
EOL ENVIRONMENT AND SURVIVAL AFTER ANTICANCER TREATMENT
patients. In addition, regarding where patients received EOL
supportive care, 118 were in the hospital group (65.2%) and
63 in the neighboring hospice/home care group (34.8%).
Analyses of intra- and postcancer
treatment survival
Figure 1 shows the PCS curve of all patients. The median
ICS and PCS were 88.1 and 13.3 weeks, respectively. On
stratifying by the maximum opioid intake, the PCS of patients
who received £ 25 mg did not significantly differ from those
who received > 25 mg (log rank: P = 0.301) (see Fig. S1).
Figures 2 and 3 show ICS and PCS curves, respectively, on
stratification by where patients received EOL supportive
care. With regard to the ICS, the curves mostly overlapped,
and we did not identify a difference between patients in the
hospice/home care cohort and those in the hospital cohort
(log rank: P = 0.6196) (see Fig.2). In contrast, patients in the
hospice/home care group showed a more significantly fa-
vorable PCS than those in the hospital group (log rank:
P = 0.029) (see Fig. 1).
Distribution of palliative treatments
We subsequently examined whether EOL medical care
differs between the two above-mentioned EOL site cohorts.
Table 2 shows the distribution of various palliative applica-
tions, including pain control, intravenous hyperalimentation,
thoracentesis/paracentesis, nasogastric/ileus tube, ureteral
stent/nephrostomy, blood transfusion, oxygen inhalation,
sedation, and colostomy/intestinal bypass. On stratification to
the two EOL site cohorts, there was no significant difference
in the distribution of all palliative treatments excluding co-
lostomy/intestinal bypass (see Table 2).
Multivariable analyses
Cox multivariable analysis was conducted to simulta-
neously examine the independent effect on PCS and elimi-
FIG. 1. Kaplan-Meier estimated survival after anticancer treatment (PCS) of all 181 patients with gynecologic malig-
nancies.
327
nate selection bias from a number of clinicopathologic
factors as thoroughly as possible. The age ( < 60 versus ‡ 60),
carcinoma type (uterine versus ovarian), pelvis invasion
(positive versus negative), maximum opioid use ( £ 25 mg or
none versus > 25 mg), and site of supportive care (hospital
versus hospice/home care) were entered into the analysis (see
Table 3). Consistent with the results of univariable analysis,
the age and EOL site retained significance as independent
prognostic factors in poor PCS in the multivariable model
(age: HR = 0.679, 95% CI, 0.496 0.928, P = 0.0151; EOL
site: HR = 0.704, 95% CI, 0.511 0.970, P = 0.0319) (see
Table 3). Therefore, patients who spend the remainder of
their lives in a hospice/their own home showed a significantly
longer PCS than those at a hospital in both uni- and multi-
variable analyses.
Discussion
Palliative care is an essential approach to cancer care that,
along with symptom control, focuses on aspects of life im-
portant to patients and their families in an attempt to protect
and relieve suffering.8,9 According to a previous report, an
earlier introduction of palliative care led to significant im-
provements in the quality of life (QOL). In the present study,
using novel prognostic parameters: ICS and PCS, we re-
vealed that patients in the hospice/home care group show a
significantly more favorable prognosis after aggressive an-
ticancer treatment than those in the university/general hos-
pital group. Although this difference may merely reflect
disease characteristics rather than an essential survival ad-
vantage in the former group, we thought that the quality of the
EOL of the former group was inferior to that of the latter.
Indeed, there was a likelihood of selection bias and treatment
heterogeneity, since our cohort included the limitations as-
sociated with any retrospective study. As such, we deter-
mined three major causes of bias: (1) duration of aggressive
treatment, (2) difference in various accompanying pallia-
tive treatments, and (3) multifactorial effect of several
328
FIG. 2. Kaplan-Meier estimated ICS of patients on stratifying by the EOL site.
clinicopathologic indicators. First, there may be a tendency
whereby the therapeutic interval between the initial to final
aggressive treatment was longer in the hospital than hospice/
home care group. In the current analysis we examined whe-
ther there was a difference in the survival duration between
the date of initial treatment and the last date of anticancer
treatment between the two groups. However, contrary to our
expectations, such intervals did not significantly differ be-
tween the two EOL site cohorts. Regarding the second pos-
sibility, we wondered whether or not hospital patients were
more likely to undergo intensive medical care and an invasive
procedure based on their inherent disease progression. If so,
the remaining oncologic outcome of those patients may be-
come biased, resulting from a poorer and more severe general
condition. In this regard we examined the distribution of
various medical treatments between the two groups, includ-
ing palliative surgeries and encompassing best supportive
FIG. 3. Kaplan-Meier estimated PCS of patients on stratifying by the EOL site.
KAJIYAMA ET AL.
care. Nevertheless, we did not identify any significant dif-
ference in the distribution pattern of most palliative inter-
ventions, even regarding the amount of opioid use. To
eliminate the third possible bias as far as possible, we con-
ducted a multivariable analysis, including the age, carcinoma
type, recurrence pattern, maximum opioid use, and EOL site.
Consequently, we identified the EOL site as an indepen-
dent prognostic factor for survival after aggressive cancer
treatment.
Despite the fact that the survival duration until the final
aggressive treatment and subsequent palliative modality were
almost the same, we identified a survival benefit in patients in
the hospice/home care group compared to the hospital group.
Connor and colleagues retrospectively analyzed the survival
of 4493 patients with congestive heart failure and five types
of malignancy, including breast, colon, lung, pancreatic, and
prostate cancer, on stratification by hospice use/nonuse. For
EOL ENVIRONMENT AND SURVIVAL AFTER ANTICANCER TREATMENT 329

Table 2. Distribution of Palliative Treatment which Patients Received

Site of palliative treatment
Palliative treatment No. Hospice/home care (N = 63) % University/general hospital (N = 118) % P-value
Intravenous hyperalimentation 38 17 27.0 21 17.8 0.148
Thoracentesis/paracentesis 36 13 20.6 23 19.5 0.854
Nasogastric/ileus tube 28 11 17.5 17 14.4 0.588
Ureteral stent/nephrostomy 21 10 15.9 11 9.3 0.019
Blood transfusion 19 5 7.9 14 11.9 0.412
Oxygen inhalation 14 2 3.2 12 10.2 0.093
Sedation 6 0 0.0 6 5.1 0.069
Colostomy/intestinal bypass 8 6 9.5 2 1.7 0.015

the six patient populations combined, the mean survival was
29 days longer for hospice than for nonhospice patients.6
Moreover, according to a previous study from Keyser and
colleagues, in a subset of 81 patients with recurrent gyne-
cologic malignancies the median overall survival for 29
nonhospice patients was 9 months versus 17 months for 52
hospice patients, and this difference was significant.10 Nei-
ther examination suggested any adverse effect on survival for
hospice patients.
In multivariate analysis of our study, the extra-hospital
care was an independent prognostic indicator. If the extra-
hospital care even partially contributes a survival benefit,
what is a possible reason? Assuming that the medical care for
terminally ill patients was the same between the two groups,
mental QOL was likely to be higher in those in an extra-
hospital environment. Indeed, a prior study reported that
QOL influences survivorship in cancer patients.11,12 Prob-
ably, mental satisfaction and peace of mind related to the
extra-hospital care may result in enrichment of patients’ lives
and may have a positive effect of improving survival, al-
though this is a hypothesis. In this context, we think that
further psychosocial factors should be considered in both
cohorts, including family accompanying status, religious
outlook, and philosophy about death. We hope that these
additional factors will be adequately evaluated in a future
prospective study. Many patients did not demand the cessa-
tion of chemotherapy, even immediately before the worsen-
ing of the general condition, either for reassurance or because
they and their oncologist believed the continuation of such
treatment to be fundamentally beneficial for their remaining
life. However, counseling patients on EOL decisions is cru-
cial for both doctors and families, although it involves
marked difficulties. Patients who undergo continuous salvage
chemotherapy in hospitals immediately before worsening of
their general condition may lose a chance to go to a hospice/
home. Indeed, the discontinuation of anticancer therapy
should be considered if adverse effects surpass the benefits.
Therefore, we think that it is important to keep an open mind
regarding such transfer and not persist with aggressive
treatment if it is harmful.
On the other hand, in the present study we identified that
the disadvantage of PCS in elderly patients compared with

Table 3. Uni- and Multivariable Analysisa of Clinicopathologic Parameters in Relation

to PostCancer Treatment Survival of Patients
Postcancer treatment survival
Univariable Multivariable
Hazard ratio (95% CI) P Hazard ratio (95% CI) P
Age
< 60 1 1
60 0.634 (0.468 0.858) 0.0032 0.679 (0.496 0.928) 0.0151
Carcinoma
Uterine 1 1
Ovarian 1.451 (1.078 1.9549) 0.0141 1.644 (1.1960 2.259) 0.0022
Pelvis invasion
Positive 1 1
Negative 0.951 (0.703 1.287) 0.7446 0.914 (0.662 1.260) 0.5823
Maximum opioid useb
25 mg or none 1 1
> 25 mg 1.167 (0.870 1.567) 0.3031 1.214 (0.895 1.647) 0.2133
Site of palliative care
University/general hospital 1 1
Hospice/home care 0.712 (0.523 0.969) 0.0307 0.704 (0.511 0.970) 0.0319
a b c
Cox Hazard Model, All intakes were converted to per os morphine. Converted dose to oral morphine intake
330
younger patients remained significant in univariate analysis.
Furthermore, age was an independent prognostic indicator in
multivariate analysis. According to prior studies, aging was
shown to be a poor prognostic factor for patients with gy-
necologic malignancy. Indeed, chemotherapy-associated
toxicities, including a febrile neutropenia, is more frequently
observed in elderly patients than in younger patients. In
general, elderly cancer patients are likely to show an age-
related decline of functional reserve in multiple organs and
the higher prevalence of comorbidities. It is likely that our
current findings reflect these unstable physical conditions
specific to elderly patients.
Limitations of our study include its retrospective nature.
The current investigation may also be biased, in that patients
were treated by different physicians, and the criteria for the
cessation of aggressive chemotherapy and subsequent intro-
duction to extra-hospital care were heterogeneous. In addi-
tion, unfortunately, our study lacked information on QOL.
QOL assessments are imperative in assessing cancer burden,
treatment, and prognosis.13 According to prior reports,
baseline QOL data have been associated with survival in
gynecologic cancer.14–16 Thus we must appropriately balance
QOL with improvement of survival. If we wish to resolve
such bias, a randomized controlled trial offers a solution;
however, it is actually very difficult to perform because of
ethical aspects and diverse environments of individual pa-
tients. At least we would like to examine these factors and
reconfirm our current results in a prospective study.
In conclusion, it is possible that the EOL environment is a
crucial prognostic factor for survival after anticancer treat-
ment. Our current data could be hypothesis generating for
gynecologic oncologists and patients confronted with this
crucial issue. To assess the further appropriateness of extra-
hospital care, we would like to accumulate more cases and
reconfirm the current results in the future.
Author Disclosure Statement
No competing financial interests exist.
References
1. Bercovitch M, Waller A, Adunsky A: High dose morphine
use in the hospice setting. A database survey of patient
characteristics and effect on life expectancy. Cancer 1999;
86:871 877.
2. Good PD, Ravenscroft PJ, Cavenagh J: Effects of opioids
and sedatives on survival in an Australian inpatient palli
ative care population. Intern Med J 2005;35:512 517.
3. Thorns A, Sykes N: Opioid use in last week of life and
implications for end of life decision making. Lancet 2000;
356:398 399.
4. Vitetta L, Kenner D, Sali A: Sedation and analgesia pre
scribing patterns in terminally ill patients at the end of life.
Am J Hosp Palliat Med 2005;22:465 473.
5. Morita T, Tsunoda J, Inoue S, Chihara S: Effects of
high dose opioids and sedatives on survival in terminally
KAJIYAMA ET AL.
ill cancer patients. J Pain Symptom Manage 2001;21:
282 289.
6. Connor SR, Pyenson B, Fitch K, Spence C, Iwasaki K:
Comparing hospice and nonhospice patient survival among
patients who die within a three year window. J Pain
Symptom Manage 2007;33:238 246.
7. Temel JS, Greer JA, Muzikansky A, et al.: Early palliative
care for patients with metastatic non small cell lung cancer.
N Engl J Med 2010;363:733 742.
8. Levy MH, Back A, Benedetti C, et al.: NCCN clinical
practice guidelines in oncology: Palliative care. J NCCN
2009;7:436 473.
9. McCartney CF, Larson DB: Quality of life in patients with
gynecologic cancer. Cancer 1987;60:2129 2136.
10. Keyser EA, Reed BG, Lowery WJ, et al.: Hospice enroll
ment for terminally ill patients with gynecologic malig
nancies: Impact on outcomes and interventions. Gynecol
Oncol 2010;118:274 277.
11. Quinten C, Coens C, Mauer M, et al.: Baseline quality of
life as a prognostic indicator of survival: A meta analysis of
individual patient data from EORTC clinical trials. Lancet
Oncol 2009;10:865 871.
12. Wenzel L, Huang HQ, Monk BJ, Rose PG, Cella D:
Quality of life comparisons in a randomized trial of interval
secondary cytoreduction in advanced ovarian carcinoma: A
Gynecologic Oncology Group study. J Clin Oncol 2005;23:
5605 5612.
13. von Gruenigen VE, Huang HQ, Gil KM, Frasure HE,
Armstrong DK, Wenzel LB: The association between
quality of life domains and overall survival in ovarian
cancer patients during adjuvant chemotherapy: A Gyneco
logic Oncology Group Study. Gynecol Oncol 2012;124:
379 382.
14. Long HJ 3rd, Monk BJ, Huang HQ, et al.: Clinical results
and quality of life analysis for the MVAC combination
(methotrexate, vinblastine, doxorubicin, and cisplatin) in
carcinoma of the uterine cervix: A Gynecologic Oncology
Group study. Gynecol Oncol 2006;100:537 543.
15. McQuellon RP, Thaler HT, Cella D, Moore DH: Quality of
life (QOL) outcomes from a randomized trial of cisplatin
versus cisplatin plus paclitaxel in advanced cervical cancer:
A Gynecologic Oncology Group study. Gynecol Oncol
2006;101:296 304.
16. Monk BJ, Huang HQ, Cella D, Long HJ 3rd; Gynecologic
Oncology Group Study: Quality of life outcomes from a
randomized phase III trial of cisplatin with or without to
potecan in advanced carcinoma of the cervix: A Gynecologic
Oncology Group Study. J Clin Oncol 2005;23:4617 4625.
Address correspondence to:
Hiroaki Kajiyama, MD, PhD
Department of Obstetrics and Gynecology
Nagoya University Graduate School of Medicine
Tsuruma-cho 65, Showa-ku
Nagoya 466-8550, Japan
E-mail: kajiyama@med.nagoya-u.ac.jp