The Role of Total Parenteral Nutrition in Patients with Peritoneal Carcinomatosis: A Systematic Review and Meta-Analysis

Simple Summary Patients with peritoneal carcinomatosis often develop complications which prevent them from receiving adequate oral intake. This can contribute to malnutrition, as well as increased morbidity and mortality. In such patients, total parenteral nutrition, which involves the intravenous administration of nutrients thereby bypassing the oral route, can be lifesaving. In this study, we performed a systematic review and meta-analysis of the existing literature to evaluate the effect of total parenteral nutrition on the survival of patients with peritoneal carcinomatosis. In light of the limited treatment options available, total parenteral nutrition may improve survival outcomes, but further studies are needed to conclude definitively. Abstract Peritoneal carcinomatosis (PC) is often associated with malnutrition and an inability to tolerate enteral feeding. Although total parenteral nutrition (TPN) can be lifesaving for patients with no other means of nutritional support, its use in the management of PC patients remains controversial. Therefore, a systematic review and meta-analysis was performed to evaluate the benefit of TPN on the overall survival of PC patients, in accordance with PRISMA guidelines. A total of 187 articles were screened; 10 were included in this review and eight were included in the meta-analysis. The pooled median overall survival of patients who received TPN was significantly higher than patients who did not receive TPN (p = 0.040). When only high-quality studies were included, a significant survival advantage was observed in PC patients receiving TPN (p < 0.001). Subgroup analysis of patients receiving chemotherapy demonstrated a significant survival benefit (p = 0.008) associated with the use of TPN. In conclusion, TPN may improve survival outcomes in PC patients. However, further studies are needed to conclude more definitively on the effect of TPN.


Introduction
Peritoneal carcinomatosis (PC) refers to the metastatic involvement of the peritoneum, typically secondary to gastrointestinal, gynecological, or other rare malignancies. It is often associated with poor prognosis, disease progression [1], and high rates of malnutrition [2], A total of 192 articles were identified in a search of six databases, including fiv cles identified from references of relevant papers (Figure 1). After removing 87 dupli 105 articles were screened and 23 were shortlisted for full-text review. A total of 10 a were included in the qualitative synthesis; eight included sufficient statistical da meta-analysis. Seven studies were retrospective, with five cohort studies [15,[39][40][41][42] and two series [43,44]. The three prospective studies included a randomized controlled trial [45], cohort study [46], and case series [47] (Table 1). All participants were diagnosed PC except for 51 with gastrointestinal malignancy included as a control for the effe PC [42]; data were not extracted for this group of patients. These studies comprised a of 1660 participants, with 620 included in the meta-analysis. Records  Full-text articles assessed for eligibility (n = 23) Full-text articles excluded, with reasons (n = 13) Studies included in qualitative synthesis (n = 10 ) Studies included in quantitative synthesis (meta-analysis) (n = 8 ) Figure 1. PRISMA flow chart [30].
Seven studies were retrospective, with five cohort studies [15,[39][40][41][42] and two case series [43,44]. The three prospective studies included a randomized controlled trial (RCT) [45], cohort study [46], and case series [47] (Table 1). All participants were diagnosed with PC except for 51 with gastrointestinal malignancy included as a control for the effects of PC [42]; data were not extracted for this group of patients. These studies comprised a total of 1660 participants, with 620 included in the meta-analysis.
Only one study involved patients who received TPN after CRS-HIPEC [46]. The remaining nine comprised patients with unresectable MBO and were divided among studies in which all patients received chemotherapy [39,43], none received chemotherapy [42,44,45], and a mix of both [15,40,41,47]. There was insufficient information to divide the third group into patients receiving versus not receiving chemotherapy.
Detailed information on TPN regimens was provided in two studies [45,47]. Two stated that nutritional support was personalized according to individual patient requirements [42,44], while six did not provide details of TPN composition [15,[39][40][41]43,46]. Duration of TPN administration was reported in four studies and, in two, ranged from a mean of 24.1 (SD 27.4) to 60.70 (SD 44.49) days [42,47]; the other two reported a median of 45 (range 9-639) days [43] and an average of 10 days postoperatively [46].
A total of 13 studies were excluded from full-text screening ( Table 3). Six [48][49][50][51][52][53] included TPN dependency only as a measure of complications or assessment of outcomes, while, in seven [24,[54][55][56][57][58][59], the relationship between TPN and outcomes data could not be determined.  [55] Relationship between TPN and outcomes data could not be determined: TPN was investigated as a factor contributing to palliative care referral Swain et al., 2018 [56] No data for overall survival; complication outcomes not related to TPN Elekonawo et al., 2019 [24] Relationship between TPN and outcome data could not be determined: "setup of study did not allow for a fair comparison of TPN vs. early enteral feeding" Kubi et al., 2020 [57] Relationship between TPN and outcomes data could not be determined: TPN and surgical complications as factors of nonhome discharge Hara et al., 2018 [58] Relationship between TPN and outcomes data could not be determined Osumi et al., 2018 [59] Relationship between TPN and outcomes data could not be determined With regard to risk of bias, four nonrandomized studies were determined to be of high quality, and the other five were determined to be of moderate quality (Table 4). There were some concerns regarding the risk of bias for the single RCT, largely due to the lack of information about the randomization process (Table 5). Table 4. Quality assessment of nonrandomized studies using the modified Newcastle-Ottawa scale (NOS).

Primary Outcome Measures: Overall Survival
All studies reported OS outcomes. However, definitions for survival differed, including time from gastrostomy tube placement to death in three [15,39,40], time from start of TPN to death in two [43,47], and time from diagnosis of terminal MBO to death in one [41]. Four did not provide a definition of OS [42,[44][45][46].
Subgroup analysis based on treatment included only studies in which either all [39,43] or no patients [42,44] received chemotherapy (Figure 2b). For studies in which all participants received chemotherapy, the pooled median OS for the group receiving TPN and not receiving TPN was 90. 66

Additional Outcome Measures: Complications, Quality of Life (QOL)
Eight studies reported complications in participants. Comparing all complications in patients with TPN versus no TPN as originally planned was not possible as none of the double-arm studies distinguished complications in these groups. As cointerventions were potential confounders, quantitative analysis was only done for complications specifically attributed to TPN by study authors [41][42][43]47]. The pooled proportion of participants who experienced complications specifically attributed to TPN was 23% (95% CI: 4-41%) (Figure 2c). Of these four studies, only one reported a median follow-up time of 89.5 (range 4-2117) days [43].
There was insufficient information to conduct statistical analysis on QOL measures, as only one study investigated QOL as measured by Karnofsky Performance Status (KPS) before and after TPN [44]; other studies included only a qualitative discussion of QOL.

Sensitivity Analysis
Sensitivity analysis was performed by repeating analyses on the four high-quality studies identified using NOS [15,[39][40][41]. As there were no significant differences in the primary outcome measures, the results were considered to be robust (Figure 2d).
Sensitivity analysis could not be done for treatment-based subgroup analyses as only one study had all patients receiving chemotherapy, while the other three had a mix of patients receiving and not receiving chemotherapy. Similarly, only one study reported complications specific to TPN.

Discussion
Clinically, patients on oncological therapy are considered distinct from patients no longer receiving treatment; TPN may improve or maintain their general condition, permitting them to receive further therapy [22], or allow waiting for gut function to be restored [25]. In the latter, TPN may prolong their lifespan by mitigating the effects of malnutrition [22]. Thus, three groups of patients who might benefit from TPN were identified: (1) patients undergoing extensive bowel work (CRS-HIPEC), (2) patients with MBO undergoing chemotherapy, and (3) patients on supportive care no longer receiving oncological therapy.
Our primary objective was to evaluate the effect of TPN versus no nutritional support on OS in patients with PC. A meta-analysis of eight eligible studies demonstrated a significant survival benefit of 17.15 days (p = 0.040) in patients with MBO who received TPN versus no TPN. When this analysis was repeated with high quality studies only, this difference persisted with a significant survival benefit of 29.17 days (p < 0.001) in patients who received TPN versus no TPN. This significant survival benefit is in agreement with existing studies citing an apparent survival benefit associated with TPN [61]. Due to variations in primary disease sites, concurrent interventions, and patient factors, further subgroup analyses were performed to explore these differences.
Tumor biology is thought to play an important role in the prognosis of patients with PC. Compared to PC of non-gynecological origin, PC from gynecological primaries, particularly ovarian cancer, is thought to be more amenable to surgical intervention [13] and chemotherapy, with a better immediate prognosis [62,63]; this advantage decreases with disease recurrence and multiple lines of chemotherapy [64,65]. In contrast, PC of gastrointestinal origin is associated with a poor prognosis [66][67][68], although recent advances in more aggressive treatments such as CRS-HIPEC have resulted in improved survival [67]. In the studies included in this review, colorectal, gastric, and pancreatic cancers [23,66,68] were the more common subtypes associated with PC. One study included patients with pseudomyxoma peritonei (PMP) from appendiceal mucinous tumors, who have much better long-term outcomes after CRS-HIPEC than patients with other gastrointestinal primaries [67,69]. However, due to the lack of information on outcomes corresponding to specific primary tumor sites within individual studies, subgroup analysis based on primary tumor sites could not be performed.
We also aimed to evaluate if patients who received chemotherapy had better survival outcomes, as a surrogate measure of treatment intent in the palliative setting. All studies comprising a mix of patients receiving and not receiving chemotherapy found that chemotherapy conferred a survival benefit [15,41,47], although, in one study, this was restricted to patients younger than 55 with no liver metastases [40]. Similarly, our study demonstrated that chemotherapy conferred a statistically significant survival benefit of 48.29 days (p < 0.001) in patients receiving TPN (Figure 2b) [45] demonstrated that TPN conferred a significant survival benefit of 39 days (mean OS) in patients who received only symptomatic treatment. Taken together, these results appear to support the finding that TPN has a positive effect on the survival of PC patients, independent of the effect of chemotherapy.
Factors other than survival outcomes may be critical in evaluating the utility of longterm TPN in patients unlikely to regain the ability to eat. QOL has increasingly become a major focus of care in cancer patients [70,71]; proponents believe that TPN is justifiable only if it can improve patients' QOL beyond merely prolonging their lives [16,47,72], as associated complications [18] and time spent on TPN infusion may detract from patients' desired use of time. Our secondary aim was, therefore, to evaluate the effect of TPN on complications and QOL.
TPN has been associated with potentially life-threatening complications such as catheter-related infections, thrombosis, liver-related complications, and metabolic imbalances [73]. In our study, catheter-related bloodstream infections, reported in 45 of 211 patients, were the most common TPN-related complication, with severe infection necessitating termination of TPN in five. Hyperbilirubinemia was reported in 10 patients. However, insufficient information on follow-up time made it difficult to compare the findings to complication rates reported in existing literature. Although TPN may have played a role in other complications, comparisons could not be made as none of the double-arm studies distinguished complications in the TPN group from those in the non-TPN group.
No studies evaluated patients' QOL before and after receiving TPN using QOL-specific questionnaires. Santarpia et al. [44] reported that, in a majority of patients, functional (KPS) and nutritional parameters stabilized or improved after a month of home PN, suggesting a positive effect on QOL. In contrast, Chouhan et al. [43] observed that high morbidity rates and time spent in the hospital were likely to negatively impact QOL. No conclusions could be drawn from two studies reporting ECOG status prior to but not post TPN [41,47].
Limited data were available for analysis in patients who underwent extensive bowel resection. The single study in this category reported outcomes of patients with PMP who received surgical resection and standard postoperative TPN, including OS and Clavien-Dindo complications [46]. However, as neither a control group for TPN nor any other studies in this category were available for comparison, no conclusions on the effect of TPN could be made. Due to the lack of median OS data, the different clinical contexts, and patient prognosis, it was not meaningful to pool results with the other studies. Even so, there may be a role for TPN in the management of these patients. Investigating the effect of differing perioperative care procedures on CRS-HIPEC patients, Elekonawo et al. [24] reported that, due to slow gastrointestinal recovery, postoperative TPN was often unavoidable, even in centers where early enteral feeding was favored. Similarly, Vashi et al. [54] demonstrated that preoperative nutrition status was associated with length of stay and OS, a finding supported by the existing literature [74]. In the absence of standardized evidence-based perioperative care procedures for CRS-HIPEC [75], further investigations would be useful in characterizing the benefit of TPN in these patients.
Qualitative analysis yielded two potential areas for investigation. Firstly, platinum analogs combined with a taxane, the current standard of care for ovarian cancer [62,65], have been shown to be less effective in patients who have received multiple lines of chemotherapy due to the development of platinum resistance [65]. Data from two studies [15,39] suggests that, for PC patients with primary ovarian cancer, chemotherapy-naïve patients with MBO on first presentation may be more likely to benefit from TPN and chemotherapy. Secondly, the effect of TPN on QOL and nutritional status is believed to be correlated with the time period of receiving TPN, i.e., the longer a patient receives TPN, the more likely they would experience improved QOL and nutritional status [45,73]. Given the poor prognosis of PC patients, prospectively identifying patients likely to have a longer OS who stand to gain a greater benefit from TPN remains a challenge [16,76,77]. Two studies suggest that BMI [42], pain, KPS, albumin, and cholinesterase levels [44] may predict survival. These findings may warrant further study.

Conclusions
In conclusion, the benefit of TPN remains closely determined by the tumor biology and baseline health status of PC patients. This systematic review and meta-analysis found a small but significant difference in survival between PC patients with MBO given TPN versus no TPN. Further subgroup and qualitative analyses suggested that this benefit persisted regardless of chemotherapy administration. However, differences in disease characteristics, patient factors, and even definitions of OS which may have affected this outcome could not be further explored due to the lack of IPD, and heterogeneity remained high in the treatment subgroup analysis. In particular, the lack of level 1 evidence from RCTs made the results highly susceptible to selection bias, with initiation of TPN largely based on physician or patient preference. However, given the complexity of decision making in PC patients, conducting randomized studies will likely be difficult, and TPN will most probably continue to be given on the basis of joint decisions by patients and physicians.
It is our view that TPN should continue to be offered as an option in patients otherwise unable to take an oral diet, provided institutions possess the necessary resources and expertise to implement it. In addition, attention to patients' preferences and QOL is critical as, once initiated, patients and caregivers may find it difficult to withhold TPN despite increasing clinical deterioration and patient discomfort [16,47,72] due to fear of death [16,78]. Given the limited prognosis of the study group, the survival benefit conferred by TPN, although modest, may be crucial for some patients and, in certain cases, well worth the potential cost. Nevertheless, continued investigations, especially in the area of complications and QOL measures, are needed to allow for better clarity in decision making. In particular, the use of TPN in CRS-HIPEC patients is clearly an underdeveloped field of study, and much more needs to be done to evaluate its role.