Nutritional conditions and PFS and OS in cancer immunotherapy: the MOUSEION-010 meta-analysis

ABSTRACT

Background

The MOUSEION-010 Meta-Analysis assessed the association between nutritional status and clinical outcomes such as Progression Free Survival (PFS) and Overall Survival (OS) among cancer patients treated with immune checkpoint inhibitors (ICIs).

Methods

Nutritional status was assessed based on the Prognostic Nutrition Index (PNI), Geriatric Nutritional Risk Index (GNRI) and Controlling Nutritional Status (CONUT) indexes. Databases consulted were: Embase, PubMed, Scopus and Web of Science.

Results

PNI and GNRI indexes did not show a significant association with both PFS and OS, while CONUT index displayed a significant difference in PFS between the two groups, in favor of the control group (Z = 4.04; p < 0.01) also without any publication bias (β= −1.27; 95% CI = [−2.13; −0.42]; p = 0.10]). The same trend was recorded in OS, too (Z = 4.24; p < 0.01). However, publication bias was present (β = 1.89; 95% CI = [1.26; 2.54]; p = 0.028]) and the numerosity of the studies did not reveal the sufficient statistical power to obtain reliable results.

Conclusion

Malnutrition could negatively impact cancer patients, especially in advanced phases. Our findings could be associated with the reduction of physical ability and daily activity performance, lower compliance with treatment protocols, and shorter survival outcomes.

1. Introduction

Dysfunctions in cell metabolism negatively impact on immune cells [1]. Additionally, malnutrition could reduce the number of immune cells and increase their functional deficiencies, specifically for T cells and their related differentiation [2,3], increasing glucose intake, consequential glycolysis for rapid growth, proliferation, and cytokine secretions [3].

The hormone and cytokine alterations in response to obesity and malnutrition are strictly associated with challenges in immune cells [1], specifically in cytokine and hormone concentrations through both paracrine effects [4], due to their proximity to adipocytes [5] and systemic and endocrine effects in circulated adipose elements [6].

Previous literature suggested that metabolites could be oncogenic in altered cell signaling and blocking cellular characterization, attaining the condition of cancer [7]. Cancer progression could represent an active ecosystem, in which the tumor microenvironment could affect the functional role of the infiltrating macrophages [8,9].

Recently, immune checkpoint inhibitors (ICIs) have shown notable efficacy in cancer patients, following the practice-changing results of several phase II and III clinical trials. However, the dark side of these agents represents the onset of immune-related adverse events (IrAEs).

Literature suggested that the prognosis of several hematological and solid tumors could be influenced by the nutritional and immunological condition of patients [10,11].

Several tools were used to assess nutritional status, including the Prognostic Nutritional Index (PNI), the Geriatric Nutritional Risk Index (GNRI), and the Controlling Nutritional Status (CONUT) scores. Literature suggested several nutritional indexes which could be associated to cancer nutrition and metabolism, such as alpha-fetoprotein, albumin, and immunoglobulin [12], prognostic nutritional index (PNI), assessed as PNI = 10 × albumin value (g/dL) + 0.005 × total lymphocyte count (per mm3) [13–15], lymphocyte-to-monocyte ratio (LMR) [16], neutrophil to-lymphocyte ratio (NLR) [17], platelet-to-lymphocyte ratio (PLR) [18], blood routine, kidney function, blood glucose, blood lipids, and blood electrolytes [19].

The PNI assessed nutritional condition according to serum albumin levels, lymphocyte, and neutrophil count in the peripheral blood [20].

In particular, the PNI index included four factors, specifically: albumin, triceps skinfold, transferrin, and skin test reactivity [20]. In 1984, Onodera changed the formula to PNI=albumin (g/L)+ 5 × absolute lymphocyte count (109/L) [21]. In 2014, a systematic review and meta-analysis including 3413 cancer patients indicated PNI as a prognostic index in several types of cancers [14], with this evidence which was also supported by a subsequent review [22].

The Geriatric Nutritional Risk Index (GNRI) assessed serum albumin level and ideal body weight [23] and was adopted among hospitalized patients to monitor nutritional conditions. The GNRI stratified patients in risk-related cut offs, like major risk (GNRI <82), moderate risk (GNRI 82 to < 92), low risk (GNRI 92 to < 98), and no risk (GNRI ≥98) of malnutrition [24–26].

The Controlling Nutritional Status (CONUT) score monitored both dimensions in cancer surgeries and drug treatments [27,28], to predict Overall Survival (OS) and hepatocellular cancer (HCC) recurrence and to study post-operative phases in pancreatic, esophageal, gastrointestinal, and orthopedic surgery [29,30].

The CONUT score was based on serum albumin concentration, peripheral lymphocyte count, and total cholesterol level, also including metabolic and inflammation-related indicators which could predict the survival of cancer patients [31–36].

However, evidence had research gaps in the association between the efficacy of ICIs and nutritional status. Thus, the MOUSEION-010 meta-analysis explored the association between nutritional status and Progression Free Survival (PFS) and Overall Survival (OS) among cancer patients treated with ICIs.

2. Materials and methods

2.1. Search strategy

The present systematic review and meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) [37]. The protocol was registered with PROSPERO id no. CRD42024589944.

Keywords according to the MeSH terminology were adopted following the PIO methodology. Search terms were mixed thanks to Boolean operators (Table 1) [37]. The databases consulted were: Embase, PubMed, Scopus and Web of Science (Supplementary File 1).

Table 1.

The PIO tool for the present systematic review and meta-analysis.

Population	Cancer Patients differentiated according to the nutritional condition assessment
Intervention	Immune checkpoint inhibitor treatments
Outcome	PFS and OS differences

2.2. Inclusion and exclusion criteria

We included studies that met the following inclusion criteria: (1) Observational studies; (2) including adult cancer patients (>18 years of age); (3) reporting nutritional-specific assessments related to PFS and OS; (4) full-text available in English.

Records included in the present analysis were all observational clinical trials and quasi-experimental studies recording nutritional differences and PFS and OS among cancer patients receiving ICIs. Thus, only Hazard Ratio (HR) and CIs 95% were collected. We excluded studies enrolling pediatric patients or healthy participants, as well as review articles, letters, editorials, and opinion articles in the present analysis.

2.3. Data extraction (selection and coding)

Initially, records were identified through a systematic database search and uploaded to a reference management software and duplicates were removed. Then, two independent reviewers (E.V. & A.R.) assessed the title and abstract of the identified studies and unsuitable reports were removed. After that, articles were uploaded, and the full text were assessed more closely for eligibility. Disagreements about whether a study should be included or not were resolved by discussion and consensus. If the disagreement remained, arbitration from another reviewer was provided (L.M.). Data collection was extracted by considering study characteristics, including participants involved and incidence of cases of IrAEs.

2.4. Selected records

During the first phase of the present systematic review and meta-analysis, a total of 224 records were identified. Among these, 150 records were removed, 67 were duplicates and 83 were excluded for other reasons, as records did not include nutritional assessment data or did not deal with cancer disease, or they were letters or editorials (n = 83). After screening, a total of 74 potential records were screened, n = 3 from Embase, n = 23 from PubMed, n = 32 from Scopus and n = 15 from Web of Science. However, 39 records were excluded as they did not meet the inclusion criteria above mentioned. Finally, the remaining 35 manuscripts were screened in the present systematic review and meta-analysis (Figure 1). Specifically, 23 studies assessed nutritional status according to the PNI score, 8 according to the GNRI and 4 to the CONUT nutritional assessment.

Figure 1.

The prisma flow diagram.

2.5. Interventions and outcomes

The systematic review and meta-analysis included all studies among cancer patients both assessing Hazard Ratio (HR) and Interval of Confidences (Cis) 95% in nutritional status and PFS and OS among cancer patients receiving ICIs.

2.6. Quality assessment and risk of bias

Publication bias was evaluated according to the “Risk Of Bias In Non-randomized Studies – of Interventions” (ROBINS-I) tool, which assessed the effects of interventions in many areas of healthcare evaluation, both highlighting and appraising strengths and weaknesses, distinguishing figures according to the nutritional score adopted (Supplementary Files 2–4) [38,39], considering quality assessments of evidence varying among “very low,” “moderate,” and “hig”.

2.7. Data analysis

Studies were assessed for quality as per protocol recommendations. The information retrieved from the final selected studies was displayed using both a narrative approach and related tables. Statistical analysis was performed using the R environment (version 4.2.3). Six separate meta-analyses were conducted. Specifically, two meta-analyses were conducted for each of the nutritional status tool (PNI, GNRI and CONUT) considering both our two outcomes, as PFS and OS. For each meta-analysis, the log hazard ratios (log(HR)) and the extreme values of the 95%CI were calculated.

To assess the consistency across studies, the I² statistic was adopted with 25%, 50%, and 75% suggesting low, moderate, and high heterogeneity degrees, respectively. However, I² should be presented and interpreted cautiously in small meta-analyses. For this reason, 95% CIs were presented in addition to the point estimate. The χ²-based Q test was also applied to look for heterogeneity of effects among studies. A significant Q value (p > 0.05) suggested the presence of significant heterogeneity between studies. Random effects meta-analysis using generic inverse variance method were implemented. Forest Plots were generated for each meta-analysis [40]. Publication bias was assessed using a contour-enhanced funnel plot and performing Egger’s t-test [41].

3. Results

3.1. Included studies

Twenty-nine studies were selected in this analysis [42–70]. Tables 2–4 report all the main outcomes for each study included in the present systematic review and meta-analysis.

Table 2.

The main characteristics of the selected studies assess nutritional status thanks to the PNI (n = 23).

Author(s)  Publication year	Cancer typology Sample Size Study design	Outcome(s)
Fang et al. [42]	223 advanced NSCLC patients Observational study	Both nutritional and inflammatory indexes were considered as better survival predictors at baseline in advanced NSCLC patients.
Guller et al. [43]	99 hNSCC stage IV patients Retrospective study	Low pretreatment nutritional condition was negatively associated with immunotherapy outcomes.
Ishiyama et al. [44]	67 mUC patients Retrospective study	PNI could be considered a helpful predictive factor for disease progression in mUC patients.
Kageyama et al. [45]	34 patients aUC Retrospective study	The post‑PNI was a predictive factor for advantaged clinical outcomes in advanced urothelial carcinoma patients.
Liu et al. [46]	123 NSCLC patients Retrospective study	PNI could be considered a predictive element in early disease progression and survival outcomes in advanced NSCLC patients.
Liu et al. [47]	151 hCC Observational study	Both PNI and GNRI were compared with inflammatory markers. The GNRI was considered the best prognostic element.
Matsuura et al. [48]	160 advanced NSCLC patients Retrospective study	Both pretreatment GNRI and PNI could be considered a potential effective predictor in survival index among advanced NSCLC patients.
Oba et al. [49]	60 MBC patients Retrospective study	PNI could be a prognostic marker for MBC patients.
Ogura et al. [50]	34 NSCLC patients Retrospective study	The PNI could be useful for predicting the outcomes of CIT.
Oku et al. [51]	218 NSCLC patients Retrospective study	The PNI could help clinicians to identify patients with better treatment outcomes.
Pan et al. [52]	268 AGC patients Observational study	PNI assessment in the pretreatment phase could stratify AGC patients according to their survival advantages.
Peng et al. [53]	102 NSCLC patients Retrospective study	PNI could be helpful to clinical outcome and irAEs.
Qi et al. [54]	53 ES-SCLC patients Prospective cohort study	Pre-treatment nutritional assessment could represent an independent prognostic factor for ES-SCLC patients.
Rejeski et al. [55]	106 large B-cell lymphoma Observational study	Obesity coukd extend CAR-T infusion in its related duration.
Sakai et al. [56]	36 RMHNSCC patients Retrospective study	Inflammatory and nutritional elements could be associated with patient prognosis after chemotherapy.
Shi et al. [57]	103 NSCLC patients at III/IV stages developing irAEs Retrospective study	Pretreatment ALC could reduce irAEs in advanced NSCLC. Low PNI scores at the pretreatment phase could be associated to high level of IL-6 and lower survival.
Shijubou et al. [58]	38 NSCLC patients Retrospective study	Weight loss negatively impacted on prognosis in NSCLC patients.
Shoji et al. [59]	102 NSCLC patients Observational study	PNI-pretreatment scores could be a predictive biomarker in ICI response among NSCLC patients.
Takahashi et al. [60]	475 NSCLC patients Retrospective study	The 3 nutritional assessments seemed to have high predictive values for postoperative complications and survival among NSCLC patients.
Vaz et al. [61]	198 with gynecological cancers Retrospective study	PNI could be a prognostic marker to predict response rates among gynecologic cancer patients.
Wang et al. [62]	273 advanced gastric patients	PNI and AGR had potential predictive powers in the immunotherapy related efficacy and prognosis among advanced gastric cancer patients.
Yi et al. [63]	75 newly diagnosed MSI-H mCRC patients	Higher AST and SII and lower PNI scores predicted worse outcomes in MSI-H among mCRC patients.
Zaitzu et al. [64]	73 lung cancer patients Retrospective study	SIS could be a helpful biomarker for the efficacy of immunotherapy.

Abbreviations: AGC, Advanced or Metastatic Gastric Cancer; AGR, albuminglobulin ratio; AST, aspartate aminotransferase; aUC, Advanced urothelial carcinoma; CIT, First-line chemoimmunotherapy; ES-SCLC, extensive-stage small-cell lung cancer; ICIs, Immune checkpoint inhibitors; irAEs, immune-related adverse events; HCC, hepatocellular carcinoma; HNSCC, head and neck squamous cell carcinomas; MBC, metastatic breast cancer; mCRC, metastatic colorectal cancer; MSI-H, microsatellite instability-high; mUC, metastatic Urothelial Carcinoma; NLR, neutrophil-to-lymphocyte ratio; NSCLC, non-small cell lung cancer; RMHNSCC, recurrent or metastatic head and neck squamous cell carcinoma; SII, systemic immune-inflammation index.

Table 3.

The main characteristics of the selected studies assess nutritional status thanks to the GNRI (n = 8).

Author(s)  Publication year	Cancer typology  Sample Size  Study design	Outcome(s)
Haas et al. [65]	162 hNSCC patients Observational study	The GNRI could represent an effective predictor for response to immunotherapy in R/M HNSCC.
Karayama et al. [66]	158 NSCLC patients Prospective, multicenter cohort study	The GNRI was a predictive index of survival and could be useful to predict the efficacy of ICIs.
Liu et al. [47]	151 hCC Observational study	Both PNI and GNRI were considered as potential biomarkers with the best prognostic value.
Liu et al. [67]	57 gastric cancer patients	Low GNRI had shorter PFS and OS. GNRI was significantly associated with survival time among gastric cancer patients receiving ICIs.
Matsuura et al. [48]	160 advanced NSCLC patients Retrospective study	Pretreatment GNRI and PNI could be a potential effective predictor for survival among advanced NSCLC patients.
Peng et al. [53]	257 advanced NSCLC patients Retrospective study	Lower GNRI scores could be a clinical trigger for nutritional support in advanced NSCLC patients.
Sonehara et al. [68]	85 NSCLC patients Retrospective study	High GNRI was associated with good outcomes among NSCLC patients treated with ICIs.
Takahashi et al. [60]	475 NSCLC patients Retrospective study	The 3 nutritional assessments had high predictive scores and could improve the postoperative outcomes among resectable NSCLC patients.

Abbreviations: GNRI, geriatric nutritional risk index; ICIs, immune checkpoint inhibitors; OS, overall survival; PFS, progression free survival; R/M HNSCC, recurrent and metastatic squamous cell carcinoma of the head and neck.

Table 4.

The main characteristics of the selected studies assess nutritional status thanks to the CONUT (n = 4).

Author(s)  Publication year	Cancer typology  Sample Size  Study design	Outcome(s)
Chang et al. [69]	69 patients with advanced esophageal cancer Retrospective study	The CONUT score can be used as an effective indicator for the prognosis of patients with esophageal cancer receiving ICI.
Chen et al. [70]	146 gastric cancer patients Retrospective study	The CONUT, as a novel immuno-nutritional biomarker, may be useful in identifying gastric cancer patients who are unlikely to benefit from ICI treatment.
Sakai et al. [56]	36 RMHNSCC patients Retrospective study	Chemotherapy following ICI is highly effective. There were no significant differences in the chemotherapy regimens. Inflammatory and nutritional factors may associate with patient prognosis after chemotherapy.
Takahashi et al. [60]	475 NSCLC patients Retrospective study	The 3 nutritional assessment methods we used were found to have high predictive values for postoperative complications and survival. Preoperative nutritional conditioning may improve the postoperative outcomes in patients with resectable NSCLC.

Abbreviations: CONUT, Controlling Nutritional Status score; ICIs, immune checkpoint inhibitors; NSCLC, non-small cell lung cancer; RMHNSCC, recurrent or metastatic head and neck squamous cell Carcinoma patients.

3.2. Studies features

Fang et al. [42] explored the association between inflammatory and nutritional indexes with prognosis by enrolling 223 cancer patients. Inflammatory indexes included neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR) and systemic immune-inflammation index (SII) and nutritional indexes included PNI, hemoglobin, albumin, lymphocyte, and platelet levels (HALP). Data recorded high pre-treatment levels with increased NLR and increase SII, predicting low PFS among advanced NSCLC patients receiving PD-1 inhibitors plus chemotherapy [42].

While inflammatory and nutritional indexes could serve as independent predictors of long-term survival, inflammatory indexes were additionally linked to treatment response. Dynamics of the explored clinical parameters were stronger than baseline values in predicting survival in advanced NSCLC patients receiving PD-1 inhibitor combined with chemotherapy as first-line [42]. Also, Guller and colleagues [43] explored the association between pretreatment nutritional stats and immunotherapy effectiveness in stage IV head and neck squamous cell carcinoma (HNSCC) patients retrospectively analyzing 99 patients with their PNI scores and body mass index (BMI). PNI scores were associated to OS, PFS and immunotherapy response [PNI (OS: HR: 0.464; 95% CI: 0.265–0.814; PFS: p = 0.007 and HR: 0.525; 95% CI: 0.341–0.808; p = 0.003)]. Additionally, multivariate analysis showed that PNI was significantly associated with OS and PFS (p = 0.011). Thus, scarce pretreatment nutritional status seemed to be associated with ineffective post-immunotherapy outcomes [43].

Ishiyama et al. [44] investigated both PNI scores and lymphocyte and albumin concentrations as potential prognostic indexes in metastatic urothelial carcinoma (mUC) patients treated with pembrolizumab. The authors reported PFS and OS as independent factors from PNI index. However, among patients at their initial disease progression, PNI could be considered as a helpful predictor for prognosis and disease progression for these patients [44].

Among mUC patients, the association between post‑PNI and OS and PFS showed a better disease management, including complete and/or partial response and stable disease (p = 0.004), also indicating a significant increase in survival time (p < 0.001). However, multivariate analysis recognized higher post‑PNI score as an independent predictive element for both OS (p < 0.001) and PFS (p < 0.001) [45].

Liu et al. [46] suggested that low PNI scores were significantly associated with early disease progression (p = 0.011). Also, in this study, it was confirmed that PNI was an independent predictor of early progression and survival outcomes in advanced NSCLC patients [46].

The emergence of ICIs has provided a new treatment option for Among HCC patients PNI, nutritional risk index (NRI), GNRI, SII, systemic inflammation response index (SIRI) and advanced lung cancer inflammation index (ALI) were assessed, and prognostic associations of these biomarkers were assessed. Associations between GNRI, PNI, BCLC stage and TumorNodeMetastasis (TNM) stage were significantly linked with PFS. On the other hand, GNRI, BCLC stage and TNM stage were also significantly connected with OS also resulting as effective noninvasive biomarkers in HCC treated with ICIs [47].

Matsuura et al. [48] showed significant differences between the low and high-GNRI or PNI groups and shorter PFS and OS. Specifically, the low-GNRI and low-PNI groups showed significantly lower PFS and OS than the high-GNRI and high-PNI counterparts. Additionally, the high-GNRI group was assessed as an independent prognostic factor for OS and PFS, and the PNI group was an independent prognostic factor for OS in advanced NSCLC patients [48].

Oba et al. [49] showed that high PNI and low NLR scores were associated with better PFS and OS. Conversely to the other studies mentioned above, PNI was considered a more effective prognostic index than NLR for OS (PNI: p = 0.0008; NLR: p = 0.14). Additionally, PNI could be considered an independent predictive factor for PFS (p = 0.0009).

Ogura et al. [50] suggested that the PFS duration was significantly linked to the NLR, C-reactive protein-albumin ratio, PNI, and advanced lung cancer inflammation index. The PNI could be considered a prognostic index of CIT in advanced NSCLC patients [50].

Oku et al. [2023] suggested a significant association between the PNI and both PFS (p = 0.0021) and OS (p < 0.0001) in NSCLC patients receiving either pembrolizumab alone or chemoimmunotherapy. The PNI seemed to be an independent prognostic factor which might help clinicians to identify patients with better treatment outcomes [51].

Pan et al. [52] showed that the low pretreatment PNI level of AGC patients was significantly associated with briefer PFS (p < 0.001) and OS (p < 0.001). During ICIs pretreatment, the PNI assessment might help to recognize AGC patients who will receive an advantage from ICI therapy [52].

Peng et al. [53] reported that NLR < 5, LDH <240 U/L, or PNI ≥ 45 were all advantaged conditions associated to better outcomes. Multivariate analysis performed confirmed that these values were all independently associated with both better PFS (p = 0.049, 0.046, 0.014) and more extended OS (p = 0.007, 0.031, <0.001). Additionally, PNI and NLR scores were also linked to the onset of irAEs [53].

Among ES-SCLC patients, NLR, PLR, lymphocyte – monocyte ratio (LMR), SII, SIRI, PNI, advanced lung cancer inflammation index (ALI), and lung immune prognostic index (LIPI) were evaluated, and pre-treatment PLR appeared to be an independent prognostic factor for ES-SCLC [54], specifically, PLR could be considered as the only independent prognostic factor for OS (p = 0.038) and PFS (p = 0.028) among ES-SCLC patients [54].

Also, in CD19-directed chimeric antigen receptor T-cell therapy (CD19.CAR-T) the obesity condition could negatively impact on modern T cell – based immunotherapies efficacy [55].

Among recurrent of metastatic head and neck squamous cell carcinoma (RMHNSCC) patients receiving ICI therapy, univariate analysis showed significant associations between serum albumin level, C-reactive protein level, PLR, NLR, LMR, SII, PNI and OS. LMR appeared the only one biomarker for predicting the prognosis of ICI treatment for RMHNSCC [56].

Low levels in pretreatment PNI scores and high level of IL-6 levels could be associated with worse survival [57]. Specifically, PNI ≤ 45 was associated with worse PFS and OS both among patients receiving ICIs only and in patients receiving ICI [57].

The significant association between weight loss and PD-L1 expressions with PFS was also confirmed in the Shijubou et al. study [58]. Specifically, a weight loss of >5% after the beginning of the ICI treatment was significantly associated with worse PFS. Thus, weight maintenance could be considered an important outcome for effective ICI treatment [58].

Shoji et al. [59] suggested that pretreatment PNI levels were significantly associated with response to ICI therapy in PFS (p = 0.0013) and OS (p = 0.0053) related outcomes. Additionally, multivariate analysis showed that PNI, C-reactive protein (CRP) and NLR were assessed as independent prognostic factors for PFS. Moreover, PNI was also considered an independent prognostic factor for OS (p = 0.0761) in ICI response in NSCLC [59].

Takahashi et al. [60] showed that resected NSCLC patients with malnutrition index exhibited significantly lower 5-year overall and recurrence-free survivals. Thus, preoperative nutritional support could advantage postoperative complications [71].

Among women suffering from gynecologic malignancies, the pretreatment PNI was associated with ICI response. Data showed that PNI could be considered as a predictive factor as severe malnutrition was linked to worse PFS (p = 0.08) and OS (p < 0.001) [61].

Low PNI index was connected to higher CEA concentrations. Additionally, low ALI levels were associated with decreased BMI levels and higher AFP concentrations. Additionally, in the non-responding ICI treatment group, PNI and AGR values were reduced, and CEA levels were higher than in the pretreatment phase. Thus, PNI and AGR were assessed as independent predictive factors in immunotherapy efficacy for advanced gastric cancer for only PFS [62].

Yi et al. [63] showed that higher PNI scores (p = 0.012) and lower AST (p = 0.049) were negative predictors of PFS, also predicting worse outcomes in MSI-H mCRC patients treated with immunotherapy [63].

The SIS and the beginning of tumor disease and PFS and OS were associated with NLR, modified Glasgow prognostic score; and PNI. However, no important results were found for the PNI in PFS and OS outcomes [64].

Considering the GNRI for the response to immunotherapy in R/M HNSCC, worse PFS was found for low GNRI (p < 0.001), and worse OS for GNRI < 92 (p < 0.001), also indicating GNRI as an independent prognostic factor for both PFS and OS [65].

Karayama et al. [66] confirmed that low GNRI was associated with significantly shorter PFS (p = 0.017) and high GNRI to significantly shorter OS (p = 0.014) and that increased GNRI was predictive of longer PFS and OS among NSCLC patients receiving nivolumab, highlighting a predictive factor to the ICI therapy efficacy [66].

Gastric cancer patients recording low GNRI scores had significantly briefer PFS (p = 0.001) and OS (p = 0.001) than those with higher GNRI patients treated with ICIs, also identifying it as an independent prognostic factor [67].

The same trend was observed in the Sonehara et al. [68] study, since high GNRI scores had significantly longer PFS ones (p = 0.041) and significantly longer median OS (p = 0.008) among NSCLC patients treated with ICIs [68].

Considering the CONUT score with SII and NLR and the prognosis, data suggested that high CONUT score was associated with a significantly worse PFS and OS in the efficacy and prognosis of ICI therapy among esophageal cancer patients [69]. Specifically, patients with high CONUT score had lower PFS (p = 0.072), and OS (p = 0.038) [70].

3.3. Meta-analysis of studies assessed nutritional status according to the PNI

3.3.1. PFS and PNI

We pooled data from 24 studies. Heterogeneity among the studies was significant (p < 0.001; tau² = 0.86 and I² = 94.8%) so a random effects model was used.

Figure 2 showed that the test for the overall effect was no significant (Z = 0.91; p = 0.36) and the Funnel Plot displayed an asymmetry (Figure 3) indicating the presence of numerous publication biases, as confirmed by the result of the Eggers test (β= −4.28; 95% CI = [−6.59; −1.97; p = 0.001]).

Figure 2.

Forest plot in PFS/PNI.

Figure 3.

Contour-enhanced funnel plot in PFS/PNI.

3.3.2. OS and PNI

There were 25 studies included in the analysis. The high heterogeneity among the studies (p < 0.001; tau² = 0.67 and I² = 96.6%) estimated a random-effects model. The results of the analysis, presented by means of the Forest Plot (Figure 4), suggested that there were no differences between the two treatments (Z = 1.83; p = 0.07).

Figure 4.

Forest plot in OS/PNI.

The symmetry displayed in the Figure 5 indicated the absence of publication bias. The results of Eggers’ test confirmed this result (β = 1.63; 95% CI = [−0.69; 3.95; p = 0.182]).

Figure 5.

Contour-enhanced funnel plot in OS/PNI.

3.4. Meta-analysis of studies assessed nutritional status according to the GNRI

3.4.1. PFS and GNRI

Data from 8 studies were considered for the analysis. The heterogeneity between the studies was very high (p < 0.001; tau² = 1.10 and I² = 95.3%), so a random-effects model was used for the analyses. The results of the analyses presented in the Supplementary File 5 indicated that there was no significant difference between the two groups (Z = −0.45; p = 0.66).

The results of the Funnel Plot (Supplementary File 6) also confirmed by Eggers’ test (β= −1.81; 95% CI = [−20.77; 17.15]; p = 0.858]), indicated that no publication bias was present. However, the low number of studies in meta-analysis could influence the statistical power to detect bias.

3.4.2. OS and GNRI

Analyses were performed from the data belonging of 8 studies. As there was high heterogeneity between the studies (p < 0.001; tau² = 1.10 and I² = 95.3%), the analyses were performed using a random-effects model. The Forest Plot (Supplementary File 7) indicated that the results of the analyses did not reveal any significant differences between the two groups (Z = −0.48; p = 0.63). The results of the Eggers test (β= −7.57; 95% CI = [−25.01; 9.86]; p = 0.427]) and the Funnel plot (Supplementary File 8) suggested that no publication bias was present. However, the low number of studies in the meta-analysis could impact on the statistical power to detect bias.

3.5. Meta-analysis of studies assessed nutritional status according to the CONUT

3.5.1. PFS and CONUT

A total of 4 studies were considered for these analyses. No heterogeneity was present between the studies (p = 0.93; tau² = 0 and I² = 0%). The results of the meta-analysis, presented in the Forest Plot (Supplementary File 9), indicated that there was a significant difference between the two groups, in favor of the control group (Z = 4.04; p < 0.01). However, both the Funnel Plot (Supplementary File 10) and the Eggers’ test (β= −1.27; 95% CI = [−2.13; −0.42]; p = 0.10]) indicated that no publication bias was present. However, the numerosity of the data did not allow sufficient statistical power to reliably determine this result.

3.5.2. OS and CONUT

Four studies were included in the analyses and no heterogeneity was detected (p = 0.86; tau² = 0 and I² = 0%). The Forest Plot results (Supplementary File 11) indicated that a significant difference was present between the two groups, in favor of the control group (Z = 4.24; p < 0.01). Even though both the Funnel Plot (Supplementary File 12) and Eggers’ test suggested the presence of publication bias (β = 1.89; 95% CI = [1.26; 2.54]; p = 0.028]). The numerosity of the studies did not allow for sufficient statistical power to obtain reliable results.

4. Discussion

An up-to-date extensive systematic review and meta-analysis of available observational studies was conducted to explore the association between nutritional status and PFS and OS among cancer patients treated with ICIs.

Literature suggested that malnutrition was directly associated to cancer death [72], due to organ dysfunctions, gut microbiota alterations and interferences in metabolic and immune disorders [73].

Thus, it seemed essential to understand what could be identified as the most complete nutritional index to better assess nutritional condition in cancer patients. In this regard, the Body Mass Index (BMI) was not recommended for this purpose as the BMI was a punctual index and seemed to not exhaustively assess the nutritional condition in cancer patients treated with ICI [74].

In the present systematic review and meta-analysis, the nutritional status was assessed according to PNI, GNRI, and CONUT indexes. Considering the PNI index and PFS, the results of the model reported that the overall effects were no significant. However, several publication biases were also recorded. The same trend was reported for OS and PNI since there were no differences between the two treatments, also considering the lack of publication bias. These meta-analyzed data were also in agreement with previous studies in which the predictive importance of nutritional status in ICI treatment and its related efficacy has not been defined, also considering the OS trend, too [71,75]. On the other hand, a recent study suggested that NSCLC patients recording a PNI ≤ 45.5 also reported significant worse PFS trend than ones with PNI > 45.5 [59]. However, literature suggested numerous factors which negatively impacted on scarce nutritional conditions and related ICI effectiveness, like hypercatabolic activity responding to malnutrition which could promote clearance of monoclonal antibodies withdrawing through the catabolic deterioration [76]. In this regard another study demonstrated a reduced OS in cachectic ICI treated patients due to the potential negative effect of increased protein turnover [77]. Additionally, the inflammatory condition was also associated with malnutrition [78,79] negatively impacting on the adaptive immune system and, at the same time, releasing proinflammatory cytokines, like interleukin-6 (IL-6) which improved systemic glucocorticoid release [78], which also negatively interfered with immune cell activity and the consequent ICI therapy inhibition, too [80].

However, previous evidence explained that nutritional interventions were positively associated with better survival ratios in cancer patients [81,82]. However, most evidence was focused on the obesity condition and less malnutrition and to earlier stages of cancer [78].

Considering PFS and the GNRI index, in our review there was no significant difference between the two groups, also without publication. The same trend was recorded between OS and the GNRI index. However, the low number of studies in meta-analysis may lack the statistical power to detect bias. In this regard, previous positive associations were highlighted between GNRI and survival rates among cancer patients considering serum albumin [83] and body weight [84]. Conversely, the current idea that normal or high body weight could be associated with survival advantages [85]. However, all these mechanisms needed to be better understood, also referring to cancer. Past evidence suggested that GNRI could be considered a useful prognostic index in cancer patients also stratifying the cancer-related risk and prognosis among these patients [65].

Considering the PFS and the CONUT index, there was a significant difference between the two groups, in favor of the control group also without any publication bias although the lack of available studies. The same trend was also recorded in OS and CONUT index, and the numerosity of the studies did not allow for sufficient statistical power to obtain reliable results. In this regard, previous studies [86,87] have just suggested that CONUT scores could be adopted to predict numerous survival rates among patients with several cancers, thanks to the inclusion of total cholesterol and serum albumin concentrations, peripheral blood lymphocyte count and further inflammation indexes recognizing the

immune function [88,89] especially for lymphocytes, macrophages and neutrophils [90,91], which could be associated with a worse prognosis.

Finally, meta-analyzed data suggested the same trend also highlighted in previous literature reporting that malnutrition could negatively impact on cancer patients, especially in those with advanced, unresectable disease [92], by reducing physical ability and daily activity performances [93], lower compliance to treatment protocols, and shorter survival outcomes [94,95].

4.1. Strengths and limitations

As no similar meta-analysis has been published, this work may add some new information to the literature. However, this paper presented some limitations. First, the included studies were all single-center studies with small sample sizes, and there were various risks of bias such as selection bias, recall bias and measurement bias, which would seriously affect the accuracy and reliability of the research results and make it difficult to draw universally applicable conclusions. However, we analyzed nutritional status, and it could be conducible to observational studies, which very often included single-center studies with small sample sizes. Additionally, our multiple meta-analysis results showed that there was a high degree of heterogeneity among studies which could indicate different patient characteristics, treatment regimens, and time points of nutritional assessment and a variety of confounding factors which might affect the results, such as the underlying diseases of patients, lifestyle, like smoking and drinking, molecular characteristics of tumors. All these issues suggested a very intricate and difficulty determining the true relationship between nutritional status and clinical outcomes. However, literature showed only these observational studies and often all these confounding variables were not contemplated or clearly stated, and we considered the nutritional condition to take a photograph of patients treated with ICIs.

Thus, we should consider these results confounding factors may have confounded the relationship between nutritional status and progression-free survival (PFS) and overall survival (OS), leading to biased results. However, future studies will introduce the figure of a dietitian who will standardize nutritional information and any clinical pathway in nutritional both assessment and treatment, too.

5. Conclusion

Several studies suggested that nutritional condition could be considered as an independent prognostic factor in several solid tumors, including lung, gastrointestinal [96], esophageal [97], and ovarian cancers [98]. Since malnutrition could be diagnosed by a phenotypic standard, based on reduced body weight with lack of muscle mass and body mass index, and an etiological assessment, such as reduced food assumption and a proinflammatory condition [77], future studies will be necessary to better explore nutrition assessment before cancer treatment to perform a prognosis on the clinical course of the disease.

Funding Statement

This paper was not funded.

Article highlights

• Considering the PNI index and PFS, the results of the model report that the overall effects were no significant (Z = 0.91; p = 0.36). However, several publication biases were also recorded (β= −4.28; 95% CI = [−6.59; −1.97; p = 0.001]). The same trend was reported for OS and PNI, since there were no differences between the two treatments (Z = 1.83; p = 0.07), also considering the lack of publication bias (β = 1.63; 95% CI = [−0.69; 3.95; p = 0.182]).

• Considering the GNRI index and PFS, in our review there was no significant difference between the two groups (Z = −0.45; p = 0.66), also without publication bias (β = −1.81; 95% CI = [−20.77; 17.15]; p = 0.858]). The same trend was recorded between OS and the GNRI index Z = −0.48; p = 0.63); (β = −7.57; 95% CI = [−25.01; 9.86]; p = 0.427])].

• Considering the CONUT index and PFS, there was a significant difference between the two groups, in favor of the control group (Z = 4.04; p < 0.01) also without any publication bias (β= −1.27; 95% CI = [−2.13; −0.42]; p = 0.10]) although the lack of available studies. The same trend was also recorded in OS and CONUT index, too (Z = 4.24; p < 0.01).

Author contributions

Conceptualization: E.V. & A.R.

Methodology: E.V., L.M., R.M.

Investigation: E.V., A.R., L.M., O.B.

Resources: A.R. & R.M.

Data curation: E.V., A.R., L.M., V.M., O.B.

Writing – original draft preparation: E.V. & A.R.

Writing – review and editing: E.V. & A.R.

Visualization and Supervision: F.M. & M.S.

All authors have read and agreed to the published version of the manuscript.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

The authors affiliated to the IRCCS Istituto Tumori “Giovanni Paolo II,” Bari are responsible for the views expressed in this article, which do not necessarily represent the Institute.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/1750743X.2025.2483656