Drainage Percutaneous Endoscopic Gastrostomy for Malignant Bowel Obstruction in Gastrointestinal Cancers: Prognosis and Implications for Timing of Palliative Intervention

Abstract

Background: Malignant bowel obstruction (MBO) is a frequent complication in patients with advanced solid tumors. Palliative relief may be achieved by the use of a drainage percutaneous endoscopic gastrostomy (dPEG) tube, although optimal timing of placement remains unknown.

Objectives: To determine median survival after diagnosis of MBO and dPEG placement, factors associated with worse survival in MBO, factors associated with receipt of dPEG, and association of timing of dPEG placement on survival.

Methods: This observational retrospective cohort study examined 439 patients with MBO on a gastrointestinal medical oncology inpatient service. Patients were characterized by age, gender, race, primary cancer type, length of stay, readmission, complications (aspiration pneumonia or bowel perforation), and receipt of dPEG. Select factors were analyzed to examine overall survival (OS) and dPEG placement.

Results: Median survival from diagnosis of first MBO was 2.5 months. Median survival after dPEG placement was 37 days. In univariate analysis, dPEG placement, complications, longer length of stay, and readmissions were significantly associated with worse OS. Receipt of dPEG was significantly associated with younger age, longer length of stay at first admission, and shorter interval to readmission. In patients who received dPEG, longer interval from MBO diagnosis to dPEG placement did not affect OS.

Conclusion: We found that prognosis following diagnosis of MBO in patients with gastrointestinal malignancies remains poor. Our data suggest that timing of dPEG placement in MBO does not affect OS and, therefore, earlier intervention with this procedure may allow earlier and prolonged palliative relief.

Introduction

Malignant bowel obstruction (MBO) is one of the most commonly encountered complications in patients with advanced cancer. Studies have shown that patients with colorectal, ovarian, and gastric cancers have a 6%–50% likelihood of developing an MBO and it usually heralds a worsening prognosis.¹ The associated symptoms of nausea, vomiting, and severe colicky abdominal pain have a profound negative impact on quality of life.² Many patients with advanced cancer and MBO are often not optimal surgical candidates due to factors such as multiple obstruction points, persistent ascites, advanced age, diffuse intraperitoneal carcinomatosis, and suboptimal functional status.³ Moreover, the prognosis remains grim with one study showing median survival following the development of MBO to be 80 days.⁴

Treatment goals are therefore palliative in many cases. Beyond pharmacological therapy, drainage percutaneous endoscopic gastrostomy (dPEG) tubes to drain gastrointestinal secretions and decompress the bowels can alleviate symptoms. Despite the palliative benefits, there are no studies to inform optimal timing of dPEG placement and this decision is largely dependent on clinical judgment and experience.

We report a descriptive study of patients on a medical gastrointestinal oncology inpatient service diagnosed with MBO to determine factors associated with overall survival (OS), patient characteristics of those who receive dPEG, and association of timing of dPEG placement with survival.

Methods

Design, population, and variables

This observational retrospective cohort study examined 439 patients on the gastrointestinal medical oncology inpatient service at Memorial Sloan Kettering Cancer Center (MSKCC) from January 1st, 2010, to December 31st, 2014. Patients were identified using the institutional data delivery service called DataLine and were included if diagnosed with their first MBO during the study date range. This study was determined to be exempt from Institutional Review Board review at MSKCC.

Outcome measurements

All patients with MBO were characterized by the following factors: age, gender, race, primary cancer type, length of stay for first hospital admission, readmission, complications (aspiration pneumonia or bowel perforation), and whether they received a dPEG. OS in patients with MBO and as well as those receiving dPEG was determined. In those who underwent dPEG placement, we examined duration from diagnosis until dPEG placement. We also examined select factors to analyze for association with OS in those with MBO, namely age, gender, race, primary cancer type, dPEG placement, complications (aspiration pneumonia or bowel perforation), length of stay, and readmission. Finally, in those receiving dPEG, we analyzed whether interval from diagnosis to dPEG placement had an effect on OS.

Data analysis and statistics

Demographic and clinical characteristics of patients who did or did not receive dPEG were compared using the chi-square test for categorical variables and the Wilcoxon rank sum test for continuous variables. OS curves were estimated using the Kaplan-Meier method. OS after discharge was evaluated for association with receipt or no receipt of dPEG and other clinical and demographic variables using Cox regression models. The follow-up was landmarked to the date of discharge so that factors such as length of stay can be considered a baseline variable. Since dPEG placement and complications may have occurred after discharge, these factors along with readmission status were analyzed as time-dependent variables. For patients who received dPEG, OS after dPEG placement was evaluated for association with the length of time between MBO diagnosis and dPEG placement. Differences were considered significant if p ≤ 0.05.

Results

With a median follow-up among survivors of 13 months (range 0.03–63), median survival from diagnosis of first MBO was 2.5 months (95% CI 2.2–3.4) (Fig. 1). Median survival after dPEG placement was 37 days (95% CI 28–44) (Fig. 2).

FIG. 1.

Kaplan-Meier survival curve for patients from diagnosis of first MBO (n = 439). There were 343 deaths. Median survival was 2.5 months (95% CI = 2.2–3.4). Median follow-up time for survivors was 13 months. MBO, malignant bowel obstruction.

FIG. 2.

Kaplan-Meier survival curve for patients with MBO who received dPEG (n = 121). There were 103 deaths. Median survival was 1.2 months (95% CI = 1.0–1.5). Median follow-up time for survivors was 0.2 months. dPEG, drainage percutaneous endoscopic gastrostomy.

Of the 439 medical oncology GI patients identified with MBO, 121 (28%) patients received a dPEG. The baseline characteristics of the two groups are summarized in Table 1. Receipt of dPEG was significantly associated with younger age (59 vs. 62, p = 0.02), longer length of stay at first MBO admission (10 vs. 5 days, p < 0.001), and shorter interval to readmission (18 vs. 36 days, p = 0.003). Compared to the no dPEG group, the dPEG group had slightly higher rates of MBO complications (11% vs. 6%, p = 0.11) and slightly lower rates of readmissions (45% vs. 56%, p = 0.06) and fewer males (48% vs. 57%, p = 0.09). The groups did not differ significantly by race or by primary cancer site.

Table 1.

Comparison of Demographic and Clinical Characteristics by dPEG and No dPEG

Characteristics	All patients N = 439 N (%)	No dPEG, N = 318 N (%)	dPEG, N = 121 N (%)	pa
Age
Median (range)	61 (18–92)	62 (18–92)	59 (19–91)	0.02
Gender				0.09
F	200 (46)	137 (43)	63 (52)
M	239 (54)	181 (57)	58 (48)
Race				0.78
White	333 (76)	244 (77)	89 (74)
Black	52 (12)	36 (11)	16 (13)
Other	54 (12)	38 (12)	16 (13)
Primary cancer				0.70
Colon and rectum	195 (44)	142 (45)	53 (44)
Pancreas	93 (21)	70 (22)	23 (19)
Stomach	50 (11)	33 (10)	17 (14)
Other	101 (23)	73 (23)	28 (23)
Length of stay for first admission, in days
Median (range)	6 (1–45)	5 (1–42)	10 (1–45)	<0.001
Readmission				0.06
No	207 (47)	141 (44)	66 (55)
Yes	232 (53)	177 (56)	55 (45)
Number of readmissions
Median (range)	2 (1–12)	2 (1–12)	1 (1–10)	0.18
Interval from discharge to first readmission
Median (range)	27 (1–1092)	36 (1–1092)	18 (1–456)	0.003
Complications of first MBO				0.11
No	406 (92)	298 (94)	108 (89)
Yes	33 (8)	20 (6)	13 (11)

^aGroups were assessed using chi-square test for categorical variables and Wilcoxon rank sum test for continuous variables. Number of patients and percentages are presented except where noted otherwise.

dPEG, drainage percutaneous endoscopic gastrostomy; MBO, malignant bowel obstruction.

In univariate analysis (Table 2), dPEG placement, having complications, longer length of stay, and readmissions was significantly associated with worse OS (p < 0.001). Primary disease in the pancreas and stomach was associated with worse survival compared to colon and rectum (p < 0.0001). Age, gender, and race were not associated with worse survival.

Table 2.

Univariate Cox Regression Analysis for Association of Factors with Overall Survival after Discharge from First MBO (N = 439)

Characteristics	Hazard ratio	95% confidence limits		p
dPEG placementa				<0.0001
No	1.0 (reference)
Yes	3.65	2.87	4.64
Complicationsa				<0.0001
No	1.0 (reference)
Yes	2.72	1.88	3.94
Length of stay (days)	1.05	1.04	1.07	<0.0001
Readmissiona				<0.0001
No	1.0 (reference)
Yes	1.83	1.43	2.34
Age (years)	1.00	0.99	1.01	0.59
Gender
F
M	1.02	0.82	1.26	0.89
Race				0.13
White	1.0 (reference)
Black	1.38	1.00	1.91
Other	0.98	0.70	1.37
Primary cancer				<0.0001
Colon and rectum	1.0 (reference)
Pancreas	2.42	1.84	3.19
Stomach	1.85	1.30	2.64
Other	1.14	0.86	1.51

^adPEG placement, complications, and readmission were analyzed as time-dependent covariates.

In those patients who did receive dPEG, the median duration from diagnosis of MBO until dPEG placement was 10 days (range 0–470 days). Among patients who received dPEG, longer interval until dPEG placement did not differentiate patients on OS after dPEG (HR 1.00, p = 0.99) (Table 3).

Table 3.

Cox Regression Analysis for Association of Factors with Overall Survival after dPEG Placement (N = 121)

Characteristic	Hazard ratio	95% confidence limits		p
Interval from diagnosis to dPEG placement (per 10 days)	1.00	0.98	1.03	0.99

Discussion

MBO is a commonly occurring, recurring, and debilitating complication in patients with malignancy and signals a poor prognosis. Our findings demonstrate a median survival of 2.5 months following diagnosis of MBO in patients with advanced gastrointestinal malignancies, which is consistent with previous studies reporting a survival of 1–3 months.⁵ Although nasogastric tubes can be used for symptom relief, this is not a recommended long-term strategy, and a more reasonable approach for an unresolving MBO is the placement of a dPEG.^6,7 We found that timing of dPEG placement (i.e., interval from diagnosis of MBO to placement of dPEG) did not have an effect on OS. Placement of a dPEG tube in most cases allows for resumption of oral intake, which may improve quality of life and allow care at home.⁸ Complete symptom relief is achieved in a majority of patients with minimal complications.⁹ Palliative efforts for this population are especially imperative as quality-of-life assessment scores in patients with malignant bowel obstruction are extremely poor at baseline.¹⁰ Therefore, it appears advantageous to place a dPEG soon after diagnosis of MBO to provide earlier symptom relief.

We found that the median survival of patients diagnosed with MBO was 2.5 months, and this prognosis of less than 6 months indicates that diagnosis of MBO alone may warrant referral to hospice. In patients who did receive a dPEG, OS was worse and may reflect current practices of dPEG placement in patients with late-stage advanced cancer, poor performance status, and large burden of symptoms. However, further investigation would be needed to ascertain the rationale behind later placement of dPEGs.

In regard to differing demographic and clinical characteristics of our patients receiving dPEG shown in Table 1, although younger age was statistically significant, the median difference of 3 years is not clinically significant. The longer length of stay is most likely attributable to continued hospitalization for dPEG placement and monitoring. Although patients with dPEG had a shorter interval to readmission, this group had fewer readmissions overall. Further investigation is needed to determine whether readmission was related to MBO or dPEG placement and whether these patients were enrolled in hospice at the time of discharge.

In determining association of factors with OS for all patients with MBO, we found that dPEG placement, having complications, longer length of hospital stay, and readmissions were all associated with worse survival. These factors may suggest a more advanced phase of disease or reflect comorbidities aside from cancer. We have previously reported comorbidities in this population and found the mean Charlson comorbidity index was 9.5, which corresponds to the highest category for mortality within a 12-month period of time, indicating numerous comorbidities in addition to metastatic cancer.¹¹ Having primary cancer in the pancreas or stomach was also associated with worse survival compared to other gastrointestinal malignancies and likely reflects the worse generic prognosis in these malignancies.

This study has several limitations. First is that the population is limited to patients with advanced gastrointestinal malignancies and our findings may not be applicable to other populations. The term MBO is used to classify both complete and partial bowel obstructions, including the small and large intestine, and care and prognosis could differ depending on the specific classification of the obstruction. Although individual chart review to determine MBO subtype was beyond the scope of this study, we hope to discern different types of MBO in future studies to determine whether there is a difference in interventions or outcomes. Moreover, we did not examine data regarding quality of life, which is important when considering palliative interventions.

In summary, our descriptive study demonstrates that there is poor median survival following the diagnosis of MBO and it is worse in patients undergoing dPEG placement. However, the timing of dPEG placement after diagnosis of MBO did not appear to affect survival. Future prospective studies will help clarify optimal timing of dPEG placement. Given wide variation in employing the same palliation strategy of dPEG placement to alleviate symptoms of MBO, we believe that earlier intervention may result in earlier palliative relief of symptoms.

Acknowledgment

We thank Meier Hsu, BA (Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, NY) for her assistance with statistical analysis.