Colorectal cancer (CRC, including anal cancer) and gastric cancer (GC) are within the top five cancer types of all estimated cancer cases and deaths worldwide. CRC and GC represent the two major types of gastrointestinal cancers, accounting for 37.8 and 21.0% of the incidence, respectively (1). As common types of gastrointestinal tumors, CRC or GC in patients often gives rise to nutritional risk or malnutrition, which is exacerbated by surgical stress (2). In some patients with GC, skeletal muscle strength and mass decrease before surgery, which causes a vicious cycle of decline in physical function and further malnutrition, resulting in shortened survival (3, 4). Pre-operative nutrition and frailty have been reported to increase the relative risk of post-operative complications by 2–4 times (5, 6).

Infectious complications are one of the most common complications after surgery and are associated with poor prognosis. Infections after surgery can significantly increase hospitalization costs, prolong the length of hospital stay, and even lead to an increase in infection-related mortality (7). Therefore, the evaluation of perioperative risk factors is of great significance for the prevention and treatment of post-operative infections. In addition to age, BMI, ASA score, diabetes, multiple underlying diseases, and other factors (8, 9), nutritional risk and malnutrition are important risk factors for infections.

To increase awareness and allow for early recognition and treatment, many types of nutritional assessments are used in clinical practice, especially via validated nutrition screening tools. For example, the NRS2002 introduced by Kondrup et al., (10) is the preferred tool for screening and assessing hospital patients. The NRS2002 was developed by the Danish Society for Parenteral and Enteral Nutrition in 2003 and was verified in an analysis that included 128 controlled clinical trials (10). It was recommended to screen nutritional risk by the Europe Society for Parenteral and Enteral Nutrition (ESPEN) Guidelines (11). The patient-generated subjective global assessment (PG-SGA) tool, mentioned in the European guidelines, was modified according to the SGA and is a frequently used nutritional assessment tool in cancer patients (11).

However, their role in predicting post-operative infections in patients with gastrointestinal cancer is unknown. It has been shown that nutritional risk and low pre-operative nutritional status in patients with GC are associated with decreased immune function and the development of complications, especially infectious complications (12). However, Pacelli et al. (13) found that pre-operative nutritional status was not correlated with the incidence and mortality of post-operative infection-related complications in patients with GC. Hsueh et al. (14) compared five nutrition assessment tools and found that the PG-SGA performed the poorest and failed to predict any post-operative complications in patients with GC.

Previous studies generally included a range of diseases with a small number of cases, which may have led to inconsistent results. To date, there have been no multi-center studies in China on the relationship between NRS2002, PG-SGA, and post-operative infections in patients with gastrointestinal cancer. Thus, the aim of this study was to assess the prognostic value of NRS2002 and PG-SGA for post-operative infections in patients with GC and CRC who underwent curative surgery.

This prospective cohort study was conducted in 18 hospitals in China between April 2017 and March 2020. The research protocol was reviewed and approved by the Ethics Committees of each institution. The National Ethics approval number is 2014ZFYJ-010. All the participants provided written informed consent. The Clinical Trial.gov identification number is NCT03115931.

A total of 4,279 patients who underwent selective operations in general surgery departments between April 2017 and March 2020 were included. A total of 1,493 GC patients and 879 CRC patients fulfilled the criteria for enrollment in this study (Figure 1). Patients (n = 81) who had pre-operative non-cancer-related infectious diseases, those who did not undergo curative surgery (n = 65), patients who refused to sign consent form (n = 22), and patients with other conditions (n = 1,739) were excluded (Figure 1). Table 1 shows the characteristics of the study population. According to the NRS2002, in GC cohort, there were 48.9% had no nutritional risk, 45.1% had nutritional risk, and 6.0% had high nutritional risk. In CRC cohort, the proportions were 36.1, 56.9, and 7.1%. According to the PG-SGA, the percentages of patients who were well-nourished, moderately malnourished and severely malnourished in GC cohort were 61.1, 34.2, and 4.7%, respectively. And in CRC cohort were 45.8, 51.3, and 2.9%. The mean age of the GC patients was 59.8 ± 10.7 years, 72.5% were male, 2.0% had hypertension, and 8.7% had diabetes. Of the CRC patients, 60.8% were male and 39.2% female, with an average age of 60.2 ± 11.4 years. A total of 2.7% had a history of hypertension and 7.7% diabetes.

The number of post-operative infections per each NRS2002 and PG-SGA category was showed in Supplementary Table S1. We observed a gradual increase in the infection rate of CRC patients with the increase of NRS2002 scores, but it was not statistically significant (8.5, 10.4, and 12.9%, P = 0.485; Supplementary Table S1). A similar result was found in PG-SGA categories in GC patients (4.4, 3.8, and 8.7%, P = 0.192; Supplementary Table S1). The rates of post-operative infections between GC and CRC were statistically different [65 (4.4%) vs. 87 (9.9%) P < 0.001, Supplementary Table S1]. And the Test of Homogeneity of Odds Ratio with the NRS2002 categories suggested that the OR values between layers were homogenous (P = 0.456). Therefore, after controlling the NRS2002 influence on stratification factor, CRC was found to be a risk factor for the occurrences of infections, with a common-OR = 2.35, 95% CI 1.68–3.30, P < 0.001. However, the tests with PG-SGA categories suggested that the OR values between layers were heterogeneous (P = 0.047). Therefore, after PG-SGA stratification, CRC was found to be a risk factor for the occurrences of infections in patients with a “B” grade, OR = 3.70, 95% CI 2.16–6.32, P < 0.001. Those with an “A” or “C” grade, diagnosis had no effect on the occurrences of infections [OR (A) = 1.46, 95% CI 0.87–2.45, P (A) = 0.147; OR (C) = 2.50, 95% CI 0.69–9.04, P (C) = 0.166; Supplementary Table S1].

Gastric cancer and CRC had the 5th (5.6%) and 3rd (10%) highest incidences among all cancers, respectively, and the 4th (7.7%) and 2nd (9.4%) for mortality, respectively, according to Global Cancer Statistics 2020 (1). Pre-operative nutritional status is associated with short-term and long-term prognosis in gastrointestinal cancer (19–21). The aim of this study was to assess the relationship between the pre-operation nutritional status and post-operation infections in patients with GC and CRC. In our study, 51.1% of the GC patients and 63.9% of the CRC patients were at nutritional risk according to the NRS2002, and 38.9 and 54.2% had moderate or severe malnutrition, respectively, according to the PG-SGA. We found that the PG-SGA B and C was a risk factor for post-operative infections in CRC. In patients with GC, NRS2002 ≥5 was a risk factor.

There are several commonly used nutritional screening tools worldwide, including the NRS2002, malnutrition screening tools (MST), and malnutrition universal screening tools (MUST). The most widely used is NRS2002 because of its low cost, easy application, and wide applicability. In contrast to the previous nutrition score, it assessed the severity of the disease to evaluate nutritional requirements. In addition, age was also taken into account. All of these features enabled the NRS2002 tool to cover a wide range of diseases in hospital, including cancer.

We found that having a high nutritional risk, defined as NRS2002 ≥5, increased the risk of post-operative infections in patients with GC, which was similar to what was reported in previous studies. Qiu et al. (22) reported that having NRS2002 ≥3 was an independent adverse prognostic factor for overall survival in their study that included 830 patients with GC. In another study that included 880 GC patients who were undergoing a gastrectomy, NRS2002 ≥3 was significantly associated with post-operative complications (P < 0.001). However, this association disappeared when the authors performed multivariate analyses adjusting for sex, age, BMI, Charlson comorbidity index, hypohemia, hypoprotein malnutrition, tumor site, laparoscopic surgery, and sarcopenia (23). However, in our study, NRS2002 ≥5 maintained consistent prediction ability in the multifactor-adjusted models. Therefore, we believe that NRS2002 ≥5 is more predictable than NRS2002 ≥3 in indicating poor prognosis of the patient underwent radical gastrectomy. In addition, in stratified analysis of NRS2002 ≥5, we observed that the risk of post-operative infections was higher in GC patients younger than 61 years and in male GC patients, and it was lower in the patients older than 61 years and in female. However, the results in the age stratification were not statistically significant. Nutritional risk evaluated by the NRS2002 was not identified as a predictor of post-operative infections in patients with CRC. The results of Wang et al. reflected the same view. They found that there were no statistically significant differences in the incidence rates of post-operative complications between patients with and without nutritional risk, according to NRS2002 score (P = 0.546) (24). In contrast to the present findings, a previous study performed by Schwegler et al. (25) suggested that the NRS2002 was successful in predicting post-operative complications, including infections, in 186 CRC patients who were undergoing surgery (OR 2.43, P = 0.004). Correspondingly, our study was a multi-center study with a large sample size, and our outcome focused on infection complications, which was the advantage of our study compared with the former.

The NRS2002 is a screening tool, which is a fast and simple method that can be used by any healthcare professional to determine whether patients need further comprehensive nutritional assessment and nutritional treatment plan. In the meanwhile, guidelines suggest that objective and quantitative assessments should be applied to patients with abnormal nutritional screening results (11). There are several commonly used nutrition assessment tools, including the PG-SGA, Mini-Nutritional Assessment (MNA), and Global Leadership Initiative on Malnutrition (GLIM). Due to its comprehensiveness and utility, the PG-SGA appears to be one of the most useful methods for the nutritional assessment of cancer patients. As an assessment tool, the PG-SGA should be conducted by nutrition professionals, which includes a comprehensive examination and evaluation of the patients' nutrition metabolism and body function for establishing a nutritional treatment plan.

We observed that the risk of post-operative infections increased in patients with CRC in the PG-SGA B group 2.19 times (95% CI: 1.34–3.57) and the PG-SGA C group 3.57 times (95% CI: 1.24–10.27). The prognostic value of the PG-SGA for post-operative infections maybe attributed to the combination of data such as unconscious weight loss, food intake, gastrointestinal symptoms, active capability, and physical examination of the patient, which were strongly associated with negative outcomes (26, 27). In multivariable logistic regression analysis, the ORs of PG-SGA B/C increased (PG-SGA B vs. A: OR 2.62, 95% CI: 1.48–4.63; PG-SGA C vs. A: OR 5.29, 95% CI: 1.64–17.10; P for trend <0.001), which means that the prognostic value of the PG-SGA for post-operative infections increased after adjusting for age, gender, smoking, hypertension, diabetes, operation type, total bilirubin, BUN, and WBC. However, when we stratified age, we found that PG-SGA lost its predictive power for people younger than 62 years. This leads us to believe that PG-SGA may be more suitable for predicting post-operative infections in elderly colorectal cancer patients. Our study also suggested that the PG-SGA was unable to predict post-operative infections in patients with GC. Similar to this study, Seo et al. (20) found that the PG-SGA did not predict the adverse events of post-operative chemotherapy in patients with gastrectomy. Esfahani et al. (28) found that there was no significant difference in the PG-SGA scores between metastatic and non-metastatic GC patients. The reason for the weakness of the PG-SGA in predicting poor prognosis of GC patients is unknown. Meanwhile, a more suitable nutritional status assessment tool for patients with GC is still needed.

This study has a few limitations that need to be considered. Although we performed nutritional screening and assessment during the first day of admission, there were no records of NRS2002 score and PG-SGA grade after that. Therefore, we cannot guarantee that the initial NRS2002 score and PG-SGA grade had been unaltered before surgery. Significant improvement in dietitians' perception of difficulty and comprehensibility of the PG-SGA can be achieved by providing short term training. However, a perceived difficulty for the physical examination still remained, which may have affected the classifying the degree of malnutrition. Another limitation is that we did not consider the impact of perioperative nutritional support therapy. In our 18 centers, the medical staff would provide nutritional support therapy for patients with nutritional risk or malnutrition in accordance with the local practices, and there is still no uniform standard. Therefore, we ignored the effects of nutrition support therapy on results in this study. In a subsequent study, we will continue to explore the relationship between nutritional support therapy and the prognosis of cancer.

In conclusion, this study showed that NRS2002 ≥5 provided good value for clinicians in the prediction of post-operative infections in patients with GC, enabling advanced interventions. On the other hand, PG-SGA B/C was a good predictor in CRC patients. Simultaneously, this article highlighted the need for a nutritional assessment tool that can better predict clinical outcomes in patients with GC.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Clinical Trial Ethics Committee, Jinling Hospital. The patients/participants provided their written informed consent to participate in this study.

XWa, LZ, and XG contributed to the conception and design of the study. LZ and SW drafted and revised the manuscript. XWa supervised the entire project and was responsible for the conception and funding. TG collected and collated data. DH contributed to the statistical analysis plan. SW contributed to the collation and analysis of the data. XWa and DH revised the manuscript for important intellectual content. LH, BL, YG, JC, DG, ZJ, YW, FG, JZ, ZX, ZC, JX, LW, JQ, GD, HH, YN, GaL, ML, HY, WZ, YZ, HQ, XWu, KW, QC, JY, YT, PZ, GJ, BO, and GuL contributed to case enrollment and data collection. All authors critically revised the manuscript, agreed to be fully accountable for ensuring the integrity and accuracy of the work, read, and approved the final manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the Research Special Fund for Public Welfare Industry of Health (201502022) and the National Natural Science Foundation of China (81700518).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.