Parenteral nutrition versus enteral nutrition after gastric cancer surgery: a systematic review and meta-analysis of randomized controlled trials Authors: Liping Yang, Lihua Yang, Ling Yang, Chao Xia, Chaonan Fei, Peibei Duan Published: 2025 ABSTRACT BackgroundGastric cancer is a major cause of cancer mortality globally, and surgery is the primary curative treatment. Postoperative malnutrition is common and can delay recovery. Providing effective nutritional support after surgery is therefore essential, but whether parenteral nutrition (PN) or enteral nutrition (EN) offers greater clinical benefit remains unclear. This study aimed to compare the effects of PN and EN on hospital stay and nutritional outcomes in gastric cancer patients after surgery.MethodsA thorough search of PubMed, Web of Science, Embase, and Cochrane Library databases was conducted following the PICOS principles to identify randomized controlled trials (RCTs) published up to August 2025. Eligible studies enrolled gastric cancer patients who received either PN or EN after surgery, reporting on length of hospital stay, nutritional markers (albumin, prealbumin, CD4/CD8 ratio), and adverse events. Two authors independently extracted data and assessed risk of bias using the Cochrane RoB 2.0 tool.ResultsNine RCTs were included, with 641 patients in the PN group and 637 in the EN group. The risk of bias among included studies was generally moderate. Length of hospital stay was longer with EN than with PN [weighted mean difference =3.45, 95% confidence interval (CI): 2.29–4.61, P<0.001]. Albumin levels were higher with EN on days 1 and 8. Prealbumin levels were higher with PN on days 1 and 8. No significant difference was detected between PN and EN regarding retinol-binding protein levels on day 8 and the CD4/CD8 ratio on day 1. Infectious complications were more frequent with EN than with PN [relative risk (RR) =1.55, 95% CI: 1.06–2.27, P=0.02]. No significant difference was observed between EN and PN concerning overall complications (RR =1.11, 95% CI: 0.89–1.40, P=0.35).ConclusionsCompared with EN, PN could shorten the hospital stay, increase the prealbumin levels, and decrease the occurrence of infectious complications, while albumin levels were higher with EN than with PN. However, given the moderate risk of bias and heterogeneity among included studies, further high-quality RCTs are needed to determine the optimal nutritional support strategy in this population. Highlight box Key findings • This meta-analysis found that parenteral nutrition (PN) in gastric cancer patients significantly shortened hospital stays and reduced infectious complications compared to enteral nutrition (EN). While prealbumin levels increased with PN, albumin levels were higher in the EN group. What is known and what is new? • Previous studies have discussed the benefits and risks of PN and EN in various patient populations. • This analysis adds new insights specific to gastric cancer patients, highlighting that while PN can lead to shorter hospital stays and lower infection rates, it may not improve overall nutritional status compared to EN, as indicated by higher albumin levels in the EN group. What is the implication, and what should change now? • The findings suggest that PN may be favored in managing surgical recovery for gastric cancer patients due to its association with shorter hospital stays and fewer infectious complications. Clinicians should consider the balance between nutritional outcomes and infection risks when choosing nutrition strategies. Future guidelines may need to emphasize individualized nutrition plans that optimize recovery while minimizing complications, especially in high-risk surgical patients. Introduction Gastric cancer remains the fifth most frequently diagnosed malignancy and the third leading cause of cancer-related mortality worldwide (1,2), with the highest incidence in Eastern Asia, Eastern Europe, and South America (1). The incidence is higher in men than in women (1,2). Despite advances in multidisciplinary management, including surgery, chemotherapy, and radiotherapy (3,4), prognosis remains poor, with a reported 5-year survival rate of patients with gastric cancer is 37% or, more specifically, 67% in localized disease, 31% in advanced stage, and 5% in metastatic disease (2). Nutritional support is a cornerstone of perioperative care for gastric cancer patients, particularly following gastrectomy and D2 lymphadenectomy (5), procedures that often compromise oral intake and predispose patients to malnutrition and related complications (6-8). Two major approaches are commonly employed: parenteral nutrition (PN) and enteral nutrition (EN). However, the optimal strategy remains controversial. Several studies suggest that PN may prolong hospital stay, increase the risk of infections and metabolic complications, and offer little advantage in nutritional recovery compared to EN (9,10). Conversely, other reports indicate that PN is equally effective and safe, especially in patients unable to tolerate EN, and may even improve immune function and clinical outcomes in certain settings (9,11-13). Similarly, while EN has been associated with fewer complications and shorter hospital stays in some trials (13), these findings are not universal, with other studies reporting no significant differences or even favoring PN under specific circumstances. These conflicting results may stem from differences in study design, patient populations, timing and composition of nutritional interventions, and definitions of clinical endpoints. Such heterogeneity complicates the interpretation of individual studies and highlights the need for a rigorous synthesis of the available evidence. Therefore, it remains essential to investigate the impact of PN versus EN in patients with gastric cancer, with the goal of clarifying their impact on perioperative outcomes and informing clinical practice. This study sought to evaluate the effects of PN versus EN in such patients. We present this article in accordance with the PRISMA reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-242/rc) (14). Methods Literature search The search strategy was elaborated following the PICOS principle (15). Inclusion and exclusion criteria The inclusion criteria were: (I) included patients with gastric cancer; (II) the intervention group received PN; (III) the control group received EN; (IV) the outcomes were the length of hospital stay, nutritional variables (albumin, prealbumin, and CD4 cells/CD8 cells ratio), and adverse events; and (V) randomized controlled trial (RCTs). The exclusion criteria were: (I) studies presented as abstracts, case reports, meta-analyses, reviews, animal studies, or protocols; (II) publications not in English; (III) inability to retrieve the full text; or (IV) absence of available data. Search strategy PubMed, Web of Science, Embase, and the Cochrane Library databases were searched for papers published from inception up to August 2025, using “Stomach Neoplasms”, “Parenteral Nutrition, Total”, as well as other relevant keywords (Table S1). The search and study selection process was done independently by two authors (Y.L.P. and Y.L.H.). Discrepancies were examined by a third reviewer (Y.L.). Data extraction and quality assessment The data retrieved included the names of the authors, year, country, patient characteristics (male percentage, age, and cancer types), study characteristics (sample size and interventions), and outcomes. The quality assessment of included RCTs were evaluated according to the Cochrane risk of bias tool (RoB 2.0 Tool) (16), by two independent investigators (** and **). Uncertainties were discussed in the reviewer team in order to achieve consensus. The RoB 2.0 tool evaluates five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Each domain was rated as “low risk”, “some concerns”, or “high risk”. Studies were judged as low risk if all domains were low risk, high risk if any domain was high risk or multiple domains raised concerns, and “some concerns” if at least one domain raised concerns but none were high risk. Statistical analysis All analyses were conducted using STATA SE 14.0 software. Relative risks (RRs), weighted mean differences (WMDs), and their corresponding 95% confidence intervals (CIs) were utilized as effect sizes. Heterogeneity among the included studies was assessed using Cochran’s Q test and the I2 statistic. An I2 value >50% indicated significant heterogeneity. When high heterogeneity was present, a random-effects model was utilized; in all other cases, a fixed-effects model was applied (17). A P value <0.05 was considered statistically significant. Subgroup analyses based on surgical methods were performed to explore potential sources of heterogeneity. Publication bias was assessed using Begg’s and Egger’s tests. In addition, sensitivity analyses were performed to identify the individual study effects on pooled results and test the reliability of results. Results Identification and selection of the eligible studies From the databases, a total of 704 records were initially identified, with 233 being excluded prior to screening. Of the remaining 471 records that underwent screening, 170 were further excluded. Subsequently, 301 reports were considered for retrieval, but 287 were discarded due to irrelevant outcomes. Among the 14 reports assessed for eligibility, five lacked available data. Ultimately, nine studies were included in this meta-analysis (Figure 1). Study selection process. Study characteristics and quality assessment The 9 studies included 641 participants in the PN group and 637 in the EN group. Six studies were performed in China (13,18-22), one in South Korea (23), one in Japan (24), and one in the USA (Table 1) (25). With regard to the analysis of the risk of bias for all-cause mortality, five studies were considered to be at low risk and four at unclear risk of bias on the randomization process (Figure 2). Table 1Characteristics of the included studiesAuthor, yearTimeCountrySample size (n)InterventionControlMale (%)Age (years)†ParticipantsOutcomeLi, 2015 (18)2009.10–2013.10China200/200PNEN52.0%/56.0%60.8±5.9/56.0±7.6Gastric cancerLength of hospital stay, nutritional variablesOkamoto, 2023 (24)2010.01–2011.12Japan12/7PNEN75.00%/85.71%71.3 (56–82)/67.8 (61–8)Gastric cancerLength of hospital stay, adverse eventsBraga, 2001 (25)–USA131/126PNEN54.20%/53.97%62.9±12.4/64.1±13.1Gastric cancerLength of hospital stay, nutritional variables, adverse eventsLi, 2012 (19)2007.12–2008.12China20/22PN + HIICEN70%/72.73%57.6±8.4/57.2±6.9Advanced gastric cancerNutritional variablesLiu, 2011 (20)2008.02–2010.08China28/28PNEN67.86%/75%70.2±8.1/74.1±9.3Gastric cancerLength of hospital stay, nutritional variables, adverse eventsWang, 2018 (13)2012.06–2016.06China63/66PNEN50.79%/51.52%48.21±6.78/48.07±7.45Gastric cancerLength of hospital stay, nutritional variables, adverse eventsLi, 2015 (21)2006.01–2013.12China136/136PNEN––Gastric cancerLength of hospital stay, nutritional variables, adverse eventsHuang, 2015 (22)2008.02–2010.03China35/35PNEN65.71%/62.96%66.1±8.1/66.7±7.2Gastrointestinal cancerLength of hospital stay, nutritional variables, adverse eventsKim, 2012 (23)2006.03–2007.07South Korea16/17PNEN81.25%/70.59%64.5 (28–82)/60 (27–74)Gastric cancerLength of hospital stay, nutritional variables, adverse events†, Age is presented as mean ± SD [or as median (range)]. EN, enteral nutrition; HIIC, hyperthermic intraoperative intraperitoneal chemotherapy; PN, parenteral nutrition. Characteristics of the included studies †, Age is presented as mean ± SD [or as median (range)]. EN, enteral nutrition; HIIC, hyperthermic intraoperative intraperitoneal chemotherapy; PN, parenteral nutrition. Quality assessment of Cochrane risk of bias tool (RoB 2.0 Tool). Primary outcome Length of hospital stay (days) Eight RCTs presented data about the effect of PN on the length of hospital stay. The results indicate that the length of hospital stay was longer with EN than PN (13,18,20-25) (WMD =3.45 days, 95% CI: 2.29–4.61, P<0.001; I2 =80.4%, Pheterogeneity<0.001) (Figure 3A). Then, a subgroup analysis of the effect of PN on the length of hospital stay was performed according to surgical method. Six studies reported a surgical method of radical gastrectomy, and two studies reported others. In both cases, there was a benefit of PN on the length of hospital stay (radical gastrectomy: WMD =3.39 days, 95% CI: 1.49–5.29, P<0.001; I2 =85.2%, Pheterogeneity<0.001; others: WMD =3.47 days, 95% CI: 2.67–4.27, P<0.001; I2 =0.00%, Pheterogeneity=0.91) (Figure 3B). Forest plot of the length of hospital stay (days). (A) Overall. (B) Subgroup analysis (radical gastrectomy and other). CI, confidence interval; WMD, weighted mean difference. Secondary outcomes Anal exhaust time (days) Four RCTs presented data about the effect of PN on the anal exhaust time. The results indicate that the anal exhaust time was longer with EN than PN (13,20-22) (WMD =1.04 days, 95% CI: 0.51–1.57, P<0.001; I2 =98.4%, Pheterogeneity<0.001) (Figure 4A). Two studies using radical gastrectomy showed a benefit of PN on the anal exhaust time (WMD =0.78 days, 95% CI: 0.39–1.16, P<0.001; I2 =87.5%, Pheterogeneity=0.005). Two studies using other surgery methods showed a benefit of PN on the anal exhaust time (WMD =1.30 days, 95% CI: 0.89–1.71, P=0.001; I2 =91.3%, Pheterogeneity=0.001) (Figure 4B). Forest plot of the anal exhaust time (days). (A) Overall. (B) Subgroup analysis (radical gastrectomy and other). CI, confidence interval; WMD, weighted mean difference. Nutritional variables Seven RCTs presented data about the effect of PN on the albumin levels and the albumin levels were comparable between PN and EN after surgery (18-23,25) (WMD =−1.50 g/L, 95% CI: −2.99 to 0.00, P=0.05; I2 =92.8%, Pheterogeneity<0.001) (Figure 5A). Five studies using radical gastrectomy showed no benefit of PN on albumin level (WMD =−2.04 g/L, 95% CI: −4.11 to 0.02, P=0.052; I2 =95.1%, Pheterogeneity<0.001). Two studies using other surgery methods also showed no benefit of PN on albumin level (WMD =−0.29 g/L, 95% CI: −1.32 to 0.73, P=0.57; I2 =0.00%, Pheterogeneity=0.54) (Figure 5B). Forest plot of the nutritional variables. (A) Overall of albumin. (B) Subgroup analysis of albumin (radical gastrectomy and other). (C) Overall of prealbumin. (D) Subgroup analysis of prealbumin (radical gastrectomy and other). (E) Overall of CD4/CD8 ratio. (F) Subgroup analysis of CD4/CD8 ratio (radical gastrectomy and other). CI, confidence interval; SMD, standardized mean difference; WMD, weighted mean difference. Six RCTs presented data about the effect of PN on the prealbumin levels and the prealbumin levels were higher with PN than EN after surgery (18,20-23,25) (WMD: −2.71 g/L, 95% CI: −4.72 to −0.70, P=0.008; I2 =99.3%, Pheterogeneity<0.001) (Figure 5C). Four studies using radical gastrectomy showed no benefit of PN on prealbumin level (WMD =−3.86 g/L, 95% CI: −8.08 to 0.35, P=0.07; I2 =99.6%, Pheterogeneity<0.001). Two studies using other surgery methods also showed no benefit of PN on prealbumin level (WMD =−0.55 g/L, 95% CI: −1.56 to 0.47, P=0.29; I2 =93.2%, Pheterogeneity<0.001) (Figure 5D). Finally, four studies presented data on the CD4/CD8 ratio and higher CD4/CD8 ratio in PN compared with EN (18,20,22,25) (WMD =−0.13, 95% CI: −0.26 to 0.00, P=0.049; I2 =73.5%, Pheterogeneity=0.01) (Figure 5E). Three studies using radical gastrectomy showed no benefit of PN on CD4/CD8 ratio (WMD =−0.12, 95% CI: −0.33 to 0.09, P=0.25; I2 =82.0%, Pheterogeneity=0.004). Another study showed a benefit of PN on CD4/CD8 ratio (WMD =−0.12, 95% CI: −0.22 to −0.02, P=0.03) (Figure 5F). Adverse events Seven RCTs reported data about post-complications, such as diarrhea, surgical site infection, blood loss, etc. (13,20-25). The results suggest that no differences in surgical complications between EN and PN [rate difference (RD) =0.07, 95% CI: −0.09 to 0.24, P=0.38; I2 =85.7%, Pheterogeneity<0.001] (Figure 6A). Five studies using radical gastrectomy showed no benefit of PN on surgical complications (RD =0.12, 95% CI: −0.10 to 0.34, P=0.30; I2 =87.2%, Pheterogeneity<0.001). Two studies showed no benefit of PN on surgical complications (RD =−0.03, 95% CI: −0.26 to 0.20, P=0.78; I2 =74.2%, Pheterogeneity=0.049) (Figure 6B). Forest plot of the anal adverse events. (A) Overall. (B) Subgroup analysis (radical gastrectomy and other). CI, confidence interval; RD, rate difference. Publication bias No publication bias was observed when considering the length of hospital stay (Begg’s test, P=0.90; Egger’s test, P=0.35), anal exhaust time (Begg’s test, P>0.99; Egger’s test, P=0.81), albumin level (Begg’s test, P=0.23; Egger’s test, P=0.17), prealbumin level (Begg’s test, P=0.26; Egger’s test, P=0.22), CD4/CD8 ratio (Begg’s test, P=0.73; Egger’s test, P=0.76), AE (Begg’s test, P=0.76; Egger’s test, P=0.46), suggesting that these results were robust. Sensitivity analysis The sensitivity analyses suggested that the length of hospital stay, anal exhaust time, nutritional variables (albumin level, prealbumin level and CD4/CD8 ratio) and adverse events were robust (Figures S1-S6). Discussion This meta-analysis of RCTs suggests that compared with EN, PN could shorten the hospital stay and anal exhaust time, and improve the prealbumin level with comparable complications. Thus, the results support the use of PN in patients with gastric cancer after surgery. The meta-analysis by Zhao et al. (26) suggested that early PN in patients with gastrointestinal cancer improved immune function, reduced inflammatory reaction, and improved recovery. In the present meta-analysis, PN was better than EN to improve prealbumin, which is a marker of nutritional status. Prealbumin has a shorter plasma half-life than albumin and is considered a more sensitive marker of nutritional status than albumin (27,28). Previous studies have demonstrated that EN is associated with fewer complications, earlier postoperative evacuation times, shorter hospital stays, lower costs, and more stable blood glucose control. EN appears to be particularly beneficial for patients with gastric cancer who also have diabetes mellitus, due to its ability to maintain more stable glucose levels (13). However, in the present meta-analysis, the length of hospital stay was shorter with PN, representing an overall better condition of the patients. Indeed, the length of hospital stay after surgery is influenced by complications, recovery, and nutritional status, among others. Still, beyond the nutritional status, PN and EN could have other advantages. Indeed, Yao et al. (29) reported that EN was better than PN for improving insulin sensitivity after gastric surgery, again supporting the use of EN in diabetic patients. Besides, no differences were observed between PN and EN regarding the delayed hypersensitivity response, polymorphonuclear cells, interleukin-2 (IL-2) receptor levels, and total lymphocytes (25). Nevertheless, although interesting, these outcomes were reported by only one study, and no meta-analysis could be performed on them. Thus, future studies are necessary to determine the effects of PN in a comprehensive manner. It is known that PN is associated with a risk of overfeeding, and it has been suggested that overfeeding during PN is associated with complications (30,31). In this meta-analysis, overfeeding as an outcome could not be analyzed because of unavailable data, but the data showed that there were no differences in overall complications between PN and EN and that the infectious complications were less frequent with PN, possibly due to a better nutritional status. It is supported by previous studies that suggested that the risks of PN are exaggerated (11,12,32). The strength of this meta-analysis is the exclusive use of data from RCTs that compared PN and EN. Still, this study has several limitations. First, the number of included studies was limited, resulting in variability in reported outcomes and potentially compromising the reliability of the findings. Second, some included studies had relatively small sample sizes, which may lead to an overestimation of treatment effects. Third, only English-language databases were searched, which may have led to omission of relevant studies published in other languages and introduced potential language bias. Fourth, some studies enrolled patients with digestive cancers other than gastric cancer, and subgroup data specific to gastric cancer were not always available, which may dilute the specificity of the results. Last, while five studies were assessed as having a low risk of bias in the randomization process, four were rated as unclear risk. Although sensitivity analyses indicated that the pooled results for both primary and secondary outcomes were robust, the inclusion of studies with unclear risk of bias may have introduced some uncertainty to the overall findings. Therefore, caution is warranted in interpreting these findings. Larger, well-designed studies focusing exclusively on gastric cancer populations are needed to validate these conclusions. Conclusions This meta-analysis demonstrated that compared with EN, PN shortens the hospital stay and anal exhaust time, and improves the prealbumin level with comparable complications. As an important method of nutrition support for several diseases, treatment of PN could be used in association with gastric cancer after surgery. Supplementary The article’s supplementary files as Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-242/prf