Outcomes of parenteral nutrition in patients with advanced cancer and malignant bowel obstruction

Abstract Background Malignant bowel obstruction (MBO) affects 3–15% of all cancer patients. In patients with advanced cancer and inoperable MBO, the average survival varies between four to nine weeks. Parenteral nutrition (PN) may improve survival in specific patient populations with malignant bowel obstruction. Aims This retrospective, single-center cohort study aimed to review individual patient outcomes on PN in the setting of advanced cancer with a diagnosis of MBO and identify clinical and laboratory markers predictive of short- and long-term survival to further highlight patients that would benefit from PN in the setting of an inoperable MBO. Results In a retrospective analysis of 68 patients receiving PN for inoperable MBO, the median survival was 142 (IQR: 63.3-239.5) days. Patients experienced a median number of two hospital readmissions (range: 0–10) and spent a median of 29 days (range: 0-105) in the hospital after starting PN. Eighteen (26.5%) patients developed a catheter-related bloodstream infection (CRBSI). A diagnosis of appendiceal cancer was identified as a predictive marker of improved survival (HR: 0.53, 95% CI: 0.29–0.92, p = 0.023). Conclusions The use of PN in the context of end-of-life cancer care is a practice that necessitates improvement. Recognizing the outcomes and patient experiences of PN utilization is essential to physicians and patients.


Introduction
Malignant bowel obstructions (MBO), de ned as bowel obstructions that inhibit physiological transit and digestion, are a common complication in patients with advanced stages of cancer (1).When unaddressed, MBO are often recurrent and can result in severe malnutrition with an impaired quality of life.In patients with advanced cancer and inoperable MBO, the average survival is approximately four to nine weeks (2).It is currently estimated that MBO affect 3-15% of all cancer patients; however, rates of MBO are higher in speci c subsets of cancer, namely gastrointestinal and gynecologic cancers (2).For example, 10-28% of patients with metastatic colorectal cancer and up to 80% of patients with gynecologic cancer subtypes suffer from MBO in their disease course (2,3).
As a clinical intervention, patients often receive intravenous parenteral nutrition (PN) in order to manage the progressive malnutrition from MBO (5).Intravenous nutrition includes the provision of macronutrients (carbohydrates, protein, and fat), micronutrients, electrolytes, and water necessary for survival.Thus, administering PN and bypassing the gastrointestinal tract may prolong survival and provide nutritional support to patients allowing for a continuation of cancer treatment, such as surgery or chemotherapy (6,7).However, the bene ts of PN for oncologic patients with advanced MBO are not fully understood.
Past studies have demonstrated that only about 30% of patients with a MBO surviving greater than three months bene t from PN with respect to improvements in quality of life (1).Furthermore, few studies have reported the overall survival of patients with advanced cancer and MBO based on different cancer subtypes or have identi ed predictive factors associated with increased survival (8,9).The evidence for whether home PN (HPN) improves survival or quality of life in individuals with MBO is very uncertain due to the limited available evidence for both outcomes (10).Therefore, the effectiveness of the utilization of PN in patients with advanced cancer and MBO is therefore debatable and requires additional data supporting its use.
To expand on the current medical literature, our aim was to retrospectively review individual patient outcomes on PN in the setting of an advanced cancer with a diagnosis of a MBO at our institution.In addition, our goal was to identify individual predictors of patient survival at baseline and early in the course of PN therapy in order to identify patients that will most likely bene t from PN use with respect to overall survival.

Participants and study settings
A retrospective review was performed including all patients 18 years of age or older with a diagnosis of advanced cancer and an admission for a MBO at our institution between 1/1/2013 and 4/1/2021.Cases of MBO were identi ed using ICD-9 and ICD-10 codes as detailed in Supplementary Table 1 and separately reviewed based on clinical, radiographic, and histologic information.
Supplementary Table 1 Diagnosis codes utilized for capture of patients from electronic data warehouse.

Clinical outcomes
The primary outcome of interest was patient survival on PN calculated between the date of PN initiation to the date of death.Identi able clinical risk factors were assessed as predictors of overall survival.Secondary outcomes of interest included identi cation of PN-related complications and number of hospital readmissions following PN initiation.

Statistical analysis
The primary outcome of interest was overall survival, and the secondary outcome was the identi cation of predictors of overall survival.Demographic data and complications data are presented as mean ± SD or median (IQR).Kaplan-Meier curves were generated for time-to-event data, that is from the date of PN initiation until the date of death.A Cox regression model was performed to examine the effects of predictor variables, namely demographic features, disease type and baseline laboratory values, on the time to death.Associations are described as hazard ratios (HR) with 95% con dence intervals (CI).
Univariate logistic regression was used to examined predictors of 30-day hospital readmission.A twosided P-value ≤ 0.05 was considered statistically signi cant.Statistical analysis was conducted using JMP ® 13.1.0(SAS Institute, Inc., Cary, NC).

Demographic characteristics
Three hundred twenty-ve patients were initially identi ed through an electronic search of medical records, among which 68 with advanced cancer and a MBO were discharged on home PN between 1/2013 and 4/2021 and met the inclusion criteria (Fig. 1).The median patient age was 57.
No other baseline clinical predictors were signi cant to predict overall survival (Table 2).

Hospital readmission
The initial length of stay in the hospital for the MBO and initiation of PN was 15 days (IQR: 9-19).The median number of hospital readmissions and number of days spent in the hospital after starting PN were 2 readmissions (IQR: 1-3) and 29.5 days (IQR: 19.3-44), respectively (Table 3).

Predictors of hospital readmission
The overall 30-day readmission rate was 52.2% (35/67).Older patients were less likely to be readmitted to the hospital (p = 0.037).There was no signi cant difference in readmission rates within 30 days of starting PN between gender, race, BMI, weight, weight loss > 10%, initial albumin value, cancer type, or receipt of additional chemotherapy (p = NS), as shown in Table 4.

Table 4
Univariate analysis of predictors of 30-day readmission.The overall 30-day readmission rate was 52.2% (35/67).Older patients were less likely to be readmitted to the hospital (p = 0.037).There was no signi cant difference in hospital readmission rates related to gender, race, weight or weight loss > 10%, initial albumin value, malignancy type, or receipt of additional chemotherapy (p = NS).

Hospice enrollment
Patients received PN for a median of 3.1 months (IQR: 1.6-5.5)before enrolling in hospice care (n = 40).
Patients spent a median of 8.5 days (IQR: 4-22) in hospice prior to death.Five patients declined hospice care and two patients established home palliative care.

Discussion
The use of PN in the context of end-of-life cancer care is a practice that necessitates improvement.As the treatment of cancer increasingly becomes management of a chronic disease process, appropriately identifying patients who will bene t from PN at the end-of-life is necessary.Therefore, our aim was to identify predictors of survival in order to further understand the bene t that patients may achieve with PN use.In addition, we focused on the complications and burden of PN therapy in the care of the patient.With respect to patient outcomes, we found a median survival of 4.7 months (IQR: 2.1-8), but also found a signi cant burden of hospital days at the end-of-life in patients pursuing use of PN for MBO.
Several international studies have similarly attempted to understand predictors of survival with the use of PN in the setting of advanced cancer.A retrospective review from Canada including 38 patients with advanced cancer receiving PN identi ed a median survival time of 162 days (5.4 months) and a Karnofsky performance score (KPS) greater than 50 at the time of PN initiation was associated with increased survival (9).In a similar study (N = 114) conducted in Poland, where gastric cancer was the most prevalent cancer type, a median survival of 89 days (2.9 months) was found (11).Sobocki et al.
found that mild (male 10.0-14.0;female 10-12.0g/dL), moderate (8.0-10.0g/dL), and severe (6.5-8.0 g/dL) anemia and hypoalbuminemia (< 2.5 g/dL) were correlated with decreased survival.Though we did not identify any predictive laboratory values, we found that a prior diagnosis of appendiceal cancer was correlated with increased survival.Consistent with the above studies, we found a median survival of 142 days in a mixed cancer population among which the most common primary tumor sites were of appendiceal (35.3%) and colorectal (22.1%) origin.
With expanding interest in using prognostic factors to estimate survival probability in cancer patients receiving PN, Mariani, et al. used the Glasgow prognostic score (GPS), KPS, and primary tumor site to construct a nomogram capable of predicting 3-month and 6-month survival (12).While this nomogram was not validated in an external population, the general features of performance status and primary tumor site likely impact overall survival (13).In our data, the primary tumor site of appendiceal cancer predicted an improved survival.Future studies, primarily in the current era with novel available therapies will be required to elucidate disease and therapy related factors impacting survival among individuals with MBO.
The primary predictor of survival in our cohort was a prior diagnosis of appendiceal cancer.This increased survival can be explained by likely several mechanistic factors, the greatest being the general indolent nature of a mucinous adenocarcinoma of the appendix.While appendiceal cancer encompasses a number of different cancer subtypes that vary histologically and by disease outcomes, as opposed to other aggressive cancer types such as, colorectal and pancreatic cancers, the metastatic spread of disease is often limited to the abdomen becoming a space occupying tumor within the abdomen as opposed to spread to other organ systems (14).The poor response of appendiceal cancer to standard chemotherapies also likely re ects differences in host response to the tumor burden (15,16).Since appendiceal cancers equated to the majority of cancer cases in this study, when evaluated separately from other cancer subtypes, median survival was 212 days in individuals with appendiceal cancer as opposed to 121 days in all other cancer subtypes.
Outside of individual patient survival, we also focused on outcomes of PN therapy in the setting of advanced cancer.Readmission after initiating PN is common.Kjeldsen et.al have found that nearly 80% of patients with incurable cancer receiving PN are readmitted into the hospital (17) We found an overall 30-day readmission rate of 52.2%.Older age was associated with a decreased likelihood of readmission within 30 days.Younger patients with metastatic cancer have been shown to experience higher readmission rates than older patients (18).Abdominal pain was the most common complication leading to hospital readmission in our study.Although symptom burden is highly variable, older patients with metastatic cancer generally report lower pain intensity (19).Age also affects clinician recommendations and individualized patient decisions (20).Older patients with metastatic disease experience higher levels of fatigue, emotional distress, loss of hope and pleasure, and independence.Furthermore, Cohen et.al found that older patients with metastatic pancreatic cancer are less likely to receive chemotherapy compared to younger patients (21).Older patients with metastatic cancer are also more likely to engage in early goals-of-care discussions, which have been associated with lower hospital readmission rates within 90-days (22).Though older age has been identi ed as an independent non-modi able risk factor associated with less hospital readmission, it is a unidimensional measure and observed differences in hospital readmission are more likely multifactorial.
Speci c to PN use, the CRBSI rate of 26% was high compared to the Canadian study that reported a line infection rate of 13% (9).Prior studies have identi ed risk factors for CRBSI, including using an intravenous port instead of a tunneled central line (23).Multiple technical explanations are plausible.
Intravenous port catheters are more challenging to place, creating a higher threshold for removal (24) and small uid reservoirs can build up if not ushed appropriately, leading to microbial growth within the catheter.Of note, intravenous ports were the most commonly used site for PN access in this cohort and likely contributed to the higher infection rate.Further assessment is needed in this cohort to identify the additional individual risk factors for developing CRBSI and minimize its impact on patient outcomes.
Lastly, we also demonstrated the burden of hospitalization on patients with a MBO on PN to include a median of 2 (IQR: 1-3) future readmissions, most commonly for worsening abdominal pain.Patients also spent a median of 29 (IQR: 19-44) days in the hospital after initial discharge on PN which accounts for nearly 21% of the patient's survival time from PN initiation.
Quality of life on long-term HPN ranges from low to acceptable (25).Moreover, patients on HPN who depend on pain medications such as opiates or benzodiazepines experience a lower quality of life (26).
Though hospital readmissions are a critical factor for de ning patients' quality of life on HPN, few evidence-based guidelines for clinical practice have been developed (27,28).Future studies will have to consider the burden of hospitalizations in this speci c patient population and its impact on quality of life.
At present, this is the largest single center study in the United States to describe the utilization of PN in the setting of an advanced cancer (n = 68).Comparable studies have been conducted in notable regions, such as Canada, Poland, and Italy, thereby prompting the need for pertinent comparisons across healthcare systems and patient demographics.Especially in the U.S. where delivery of PN can be fragmented, this large-scale collection allowed us to appropriately describe outcomes and burdens of PN therapy.However, given the heterogenous nature of the cancer subtypes included, speci c predictors of survival outside of appendiceal cancer were not able to be determined and risk factors for hospital readmission also did not achieve statistical signi cance.Due to advancements and availability of chemotherapy and immunotherapy, identifying predictors of survival may prove to be a more complex analysis due to better outcomes with continuity of advanced treatment.Thus, gaining a comprehensive understanding of how innovative therapies in uence treatment outcomes and overall survival becomes of paramount importance.The primary limitation of this study in addition to study size is the retrospective nature of the study which limited our ability to assess patient and family caregiver burden of PN utilization.Notably, the acquisition of hospice referral data relied upon the hospital's electronic medical records.It is essential to acknowledge that certain instances of hospice referrals might have transpired without explicit documentation in the patients' charts.When attempting to determine appropriateness of utilization, future studies will require prospective patient engagement and enrollment to determine patient and caregiver perspectives on the utilization of PN.
In summary, we show that patients with appendiceal cancers experienced greater survival than patients with non-appendiceal cancers.PN-associated complications occurred in many patients, notably, catheter related blood stream infections occurred in 18 (26%) patients.Patients survived an average of 142 days but also experienced an average of 2 readmissions and 29 days in the hospital after starting PN.Future work is needed to better characterize the patient experience on PN in cases of advanced cancer.

Figure 1
Figures

Table 1
Overview of patient demographics for the study cohort.The table includes information on age, gender distribution, oncologic history, Eastern Cooperative Oncology Group (ECOG) scores, relevant laboratory values, treatment features, and survival.5%) were assessed as ECOG 3 (capable of only limited self-care, con ned to bed or chair more than 50% of waking hours).One patient (1.5%) was characterized as ECOG 4 (completely disabled; cannot carry on any self-care; totally con ned to bed or chair).Twenty-nine (46%) patients had a weight loss of greater than 10% in the six months before starting PN and 18 (25.7%)patients had a BMI of < 20.

Table 2
Cox regression time to event analysis to identify univariate predictors of overall survival.Patients with appendiceal cancer had statistically signi cant greater overall survival compared to patients with non-appendiceal cancers, demonstrating a reduced hazard ratio of 0.53 (95% CI: 0.29-0.92,p = .02)for death.No other baseline clinical predictors were signi cant to predict overall survival.

Table 3
Complications and hospital readmissions associated with PN use in advanced cancer patients.The median number of hospital readmissions and number of days spent in the hospital after starting PN were 2 readmissions (IQR: 1-3) and 29.5 days (IQR: 19.3-44), respectively.Fiftyone (75%) patients were subsequently admitted to the hospital for abdominal pain and 25 (36.8%)patients had hospitalizations for sepsis/bacteremia.Speci c to PN utilization, 18 (26.5%)had a catheter related bloodstream infection (CRSBI).
As a percent of survival time from the initiation of PN, patients spent 20.8% of their remaining time alive in the hospital.