Prognostic value of inflammation-immunity-nutrition score in patients with cancers

Abstract

Background

The inflammation-immune-nutrition score (IINS) plays a significant role in predicting survival outcomes across various cancers. The study aimed to comprehensively evaluate the prognostic value of IINS in different cancers through a meta-analysis.

Methods

A systematic literature search was conducted in PubMed, Embase, Web of Science and Scopus databases for eligible studies published up to March 15, 2026. Pooled hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated to assess the associations with overall survival (OS), disease-free survival (DFS), progression-free survival (PFS) and recurrence-free survival (RFS).

Results

A total of 10 studies from 8 articles encompassing 3979 patients were included in the analysis. Pooled analysis revealed that high IINS was significantly associated with poor OS (HR:2.54, 95% CI:1.84–3.50), PFS (HR:1.87,95% CI:1.06–3.31), DFS (HR:2.30,95% CI: 1.84–2.89) and RFS (HR:3.50,95% CI: 2.06–5.93). Subgroup analysis further demonstrated the strong prognostic value of high IINS for hepatocellular carcinoma and endometrial cancer.

Conclusions

The IINS could serve as an effective prognostic indicator for patients with cancers.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-05301-9.

Introduction

Cancer continues to represent a significant global health challenge, with the World Health Organization estimating around 20 million new cases and 10 million cancer-related deaths each year [1]. Despite notable advancements in therapy over recent decades, particularly in immunotherapy, targeted therapy, and precision medicine, significant disparities remain in treatment outcomes across different cancer types and stages. For example, while the 5-year survival rate for localized prostate cancer exceeds 99% [2], aggressive cancers like pancreatic and small cell lung cancer still exhibit dismal survival rates below 13% and 5%, respectively, even with optimal treatment [3, 4]. These stark contrasts highlight the urgent need for reliable prognostic tools capable of accurately stratifying patient risk and guiding personalized treatment strategies.

The prognosis of cancer patients is influenced by the complex interplay among systemic inflammation, immune status, and nutritional condition [5–7]. Despite significant advances in cancer therapy, accurate prognostic stratification remains challenging. Given the limitations of conventional prognostic markers, composite biomarkers that integrate inflammation, immunity, and nutrition have attracted increasing attention for their potential to more accurately stratify patient risk.

As an emerging prognostic biomarker, the inflammation-immunity-nutrition score (IINS) is calculated from high-sensitivity C-reactive protein (hs-CRP), lymphocytes, and albumin (ALB). It is believed that IINS can effectively reflect the inflammation, immunity and nutritional status of cancer patients [8]. Compared with other inflammation indices, such as prognostic nutritional index (PNI), neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), controlling nutritional status score (CONUT) and systemic immune-inflammation index (SII), IINS offers several advantages [9–12]. Its multidimensional structure accounts for both immune and nutritional status in addition to inflammation, potentially enhancing prognostic accuracy across diverse malignancies. In addition, IINS incorporates hs-CRP, a sensitive acute-phase reactant directly regulated by pro-inflammatory cytokines, which may allow more dynamic and clinically responsive risk stratification. Unlike single-component indices, IINS may be less susceptible to fluctuation of an individual parameter. While several studies have evaluated the prognostic value of the IINS in individual cancers, no comprehensive analysis has determined its prognostic performance in different malignancies [13–16]. Therefore, a systematic meta-analysis was necessary to consolidate current evidence and quantify the association between IINS and survival outcomes in cancer patients.

Materials and methods

Search strategy

We systematically searched PubMed, Embase, Web of Science, and Scopus databases up to March 15, 2026, to identify studies assessing the prognostic value of IINS in cancer patients. In addition, Google Scholar was additionally searched to identify grey literature. “Inflammation-immunity-nutrition score” was used as the primary search term, with no restrictions on language. The search strategy was presented in supplementary table S1. We screened titles, abstracts, full texts and possible references to identify qualified studies. This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Inclusion and exclusion criteria

The literature retrieval was independently conducted by three researchers. The inclusion criteria were as follows: (1) studies assessed the association between IINS and the survival outcomes of cancer patients; (2) studies provided sufficient data to calculate the hazard ratio (HR) and its 95% confidence interval (CI). The exclusion criteria were: (1) studies with incomplete data or missing key information; (2) non-original research, including case reports, reviews, conference abstracts, and commentaries.

Data extraction and quality assessment

Data were independently extracted by two investigators, and any discrepancies were resolved through discussion or consultation with the third person. We systematically extracted the following key data from the final included studies: first author, publication year, follow-up time, country, sample size, treatment methods, analysis types and survival outcomes. When HRs were directly reported, data were preferentially used. For studies without directly reported HRs, survival data was extracted from Kaplan-Meier curves using Engauge Digitizer version 4.1. HRs and 95% CIs were calculated using the methodology described by Tierney et al. and other studies [17–19]. Results from multivariate analyses were prioritized, as they more effectively control potential confounding factors. The methodological quality of all studies was assessed using the Newcastle–Ottawa Scale (NOS) [20]. Studies with an NOS score greater than 5 were considered to be of high quality and were included in the final analysis.

Statistical analysis

All statistical analyses were performed using STATA version 12.0 (StataCorp, College Station, TX, USA). Heterogeneity was assessed using the Cochrane Q test and I² statistic, and model selection considered both statistical and clinical heterogeneity. Random-effects models were applied for substantial heterogeneity, while fixed-effects models were used when studies were comparable [21, 22]. Subgroup analyses and meta-regression explored sources of heterogeneity, and sensitivity analyses confirmed the robustness of the meta-analysis. Publication bias was evaluated using Begg’s test and Egger’s test, with the trim-and-fill method applied if publication bias was detected [23, 24]. P value < 0.05 was considered statistically significant.

Results

Search results

Through a systematic literature search, we initially identified 111 relevant studies from major academic databases. After rigorous duplication, 32 duplicates were removed, leaving 79 publications with independent research value. We then conducted a preliminary screening of the titles and abstracts of these 79 studies based on predefined inclusion and exclusion criteria. Of these, 71 articles were excluded for not meeting the research topic or methodological requirements. Finally, 8 articles comprising 10 independent studies published between 2021 and 2025 were included [8, 13–16, 25–27]. Figure 1 visually illustrated the complete process from initial screening to final inclusion.

Fig. 1.

Flow diagram for study screening and selection processes

Studies characteristics

Table 1 summarized the baseline characteristics of the included studies. All studies reported overall survival(OS) data, and seven studies analyzed disease-free survival (DFS)/progression-free survival (PFS)/recurrence-free survival (RFS) data. The studies encompassed six different tumor types, including lung cancer, hepatocellular carcinoma, colorectal cancer, breast cancer, endometrial cancer and lymphoma. The NOS scores for the included studies ranged from 6 to 8, indicating that all studies were considered high quality. The specific NOS scores were shown in Table 2.

Table 1.

Basic information of the included articles

Study	Year	Country	Study type	Cancer type	Sample	Treatment methods	Follow-up time (months)	Analysis type	Survival analysis	NOS score
Cheng	2025 A	China	R	LC	302	Surgery	NA	MVA	OS	7
Cheng	2025B	China	R	LC	82	Surgery	NA	UVA	OS	6
He	2024	China	R	Lymphoma	304	Non-surgery	Median 51	MVA	OS	7
Liang	2022	China	R	HCC	204	Surgery	Median17 ± 10	MVA	OS, PFS	7
Zhang	2022	China	R	HCC	101	Non-surgery	Median11	MVA	OS, PFS	7
Li	2021	China	R	CRC	719	Surgery	Median 40	MVA	OS, DFS	8
Song	2022	China	R	HCC	801	Surgery	NA	MVA	OS, DFS	8
Jiang	2022 A	China	R	EC	775	Surgery	Median 51	UVA	OS, RFS	8
Jiang	2022B	China	R	EC	491	Surgery	Median 50	UVA	OS, RFS	7
Wang	2025	China	R	BC	200	Surgery	Median46	MVA	OS, PFS	7

R, retrospective; OS, overall survival; DFS, disease-free survival; PFS, progression-free survival; RFS, recurrence free survival; MVA: multivariate analysis; UVA: univariate analysis; NOS score, Newcastle-Ottawa Scale score; LC, lung cancer; HCC, hepatocellular carcinoma; CRC, colorectal cancer; EC, endometrial cancer; BC, breast cancer; NA, not available;

Table 2.

Newcastle-Ottawa quality assessment scale

First author	Year	Selection	Comparability	Outcome	Total
He	2024	★★	★★	★★★	7
Cheng	2025 A	★★	★★	★★	6
Cheng	2025B	★★	★★	★★★	7
Liang	2022	★★	★★	★★★	7
Zhang	2022	★★	★★	★★★	7
Li	2021	★★	★★★	★★★	8
Song	2022	★★	★★★	★★★	8
Jiang	2022 A	★★	★★★	★★	7
Jiang	2022B	★★	★★★	★★	7
Wang	2025	★★	★★	★★★	7

Association between high IINS and OS

A systematic analysis of 10 studies revealed that a high IINS score in cancer patients was significantly associated with overall OS. Because substantial heterogeneity existed among the studies (I²=79.9%), we used a random-effects model to combine the data and make the results more robust. The analysis indicated that patients with high IINS scores had a poorer survival prognosis, with the HR of 2.54 (95% CI: 1.84–3.50) (Fig. 2).

Fig. 2.

Forest plot of the association between high IINS and OS

Subgroup analysis and meta-regression for OS

We further conducted subgroup analysis and meta-regression based on cancer type, sample size, treatment method and analysis type (Table 3). Subgroup analyses confirmed the consistent prognostic value of high IINS across most subgroups. In treatment-based stratification, surgical patients showed a significant association (HR: 2.73, 95% CI: 1.88–3.98), whereas non-surgical patients demonstrated a non-significant trend (HR: 1.88, 95% CI: 0.80–4.42). Analysis by cancer type revealed particularly strong associations in hepatocellular carcinoma (HR: 3.80, 95% CI: 2.65–5.45) and endometrial cancer (HR: 3.13, 95% CI: 1.93–5.06), while lung cancer did not reach statistical significance. Both small-scale (< 300 patients; HR: 2.72, 95% CI:1.98–3.73) and large-scale (≥ 300 patients; HR: 2.42, 95% CI: 1.58–3.68) studies showed significant associations. The prognostic value was consistent in both multivariate analysis (HR: 2.36, 95% CI: 1.62–3.46) and univariate analysis (HR: 3.16, 95% CI: 2.07–4.83). Subgroup analysis showed that cancer type, sample size, treatment method and analysis type might be sources of heterogeneity. Meta-regression indicated that cancer type may be the main source of heterogeneity (P = 0.041).

Table 3.

Subgroup analysis and meta-regression for OS

Factors	Studies	HR (95%)	P	heterogeneity	P	Meta-regression	Adj R (%)	P
				I2		Tau2
Treatment method
Non-surgery	2	1.88 (0.802–4.421)	0.146	51.9	0.149	0.1227	10.39	0.288
Surgery	8	2.73 (1.88–3.98)	<0.01	82.1	<0.01
Sample size						0.1466	− 7.12	0.582
<300	4	2.72 (1.98–3.73)	<0.01	0	0.912
>300	6	2.42 (1.58–3.68)	<0.01	87.4	<0.01
Cancer type						0.01	65.2	0.041
LC	2	1.9 (0.84–4.33)	0.122	71.9	0.059
HCC	3	3.80 (2.65–5.45)	<0.01	0	0.687
EC	2	3.13 (1.93–5.06)	<0.01	0	0.802
BC	1
Lymphoma	1
CRC	1
Analysis type						0.134	1.51	0.398
MVA	7	2.36 (1.62–3.46)	<0.01	84.8	<0.01
UVA	3	3.16 (2.07–4.83)	<0.01	0	0.964

MVA: multivariate analysis; UVA: univariate analysis: LC, lung cancer; HCC, hepatocellular carcinoma; CRC, colorectal cancer; EC, endometrial cancer; BC, breast cancer;

Association between high IINS and DFS/PFS/RFS

A total of 7 studies evaluated the association between high IINS and DFS/RFS/PFS. The pooled analysis indicated the significant association between high IINS and worse DFS/RFS/ PFS (HR: 2.15, 95% CI: 1.85–2.49), with low heterogeneity across studies (I² =25.1%) (Fig. 3). In addition, the meta-analysis demonstrated that high IINS was significantly associated with inferior survival outcomes across PFS, DFS, and RFS endpoints. Specifically, the pooled HR was 1.87 (95% CI: 1.06–3.31) for PFS, 2.30 (95% CI: 1.84–2.89) for DFS, and 3.50 (95% CI: 2.06–5.93) for RFS.

Fig. 3.

Forest plot of the association between high IINS and DFS/PFS/RFS

Sensitivity analysis

To evaluate the stability of the meta-analysis results, we recalculated the combined effect size by sequentially excluding each study to assess the impact of individual studies on the overall outcome. The results indicated that excluding any single study did not substantially affect the numerical range or statistical significance of HR. These findings were visually presented in Fig. 4A, B, demonstrating the robustness of the meta-analysis results.

Fig. 4.

Funnel plot of sensitivity analysis. A sensitivity analysis for OS. B sensitivity analysis for DFS/PFS/RFS

Publication bias

We systematically assessed the potential impact of publication bias using the Begg’s test and Egger’s test. In the OS analysis, the result of the Begg’s test was not statistically significant (P = 1), while the Egger’s test was statistically significant (P = 0.030), suggesting the presence of some publication bias (Fig. 5A). To evaluate the robustness of the results, we applied the trim-and-fill method for correction analysis. The findings demonstrated that the pooled analysis remained significant (HR: 2.309, 95% CI: 1.729–3.082), confirming that the conclusion was not substantially affected by publication bias (Fig. 5B). In the analysis of DFS/PFS/RFS, the result of the Begg’s test was not statistically significant (P = 0.368), while the Egger’s test was statistically significant (P = 0.045) (Fig. 5C). Using the same adjustment method, the pooled HR remained statistically significant (HR:2.077, 95% CI: 1.691–2.551) (Fig. 5D).

Fig. 5.

Publication bias. A publication bias for OS. B trim-and-fill method for OS. C publication bias for DFS/PFS/RFS. D publication bias for DFS/PFS/RFS

Discussion

To the best of our knowledge, this study was the first comprehensive analysis of the prognostic impact of IINS in patients with cancers. The pooled results demonstrated that high IINS was significantly associated with unfavorable OS, DFS, PFS and RFS in cancer patients. Notably, the prognostic impact appeared particularly pronounced in hepatocellular carcinoma and endometrial cancer, suggesting potential tumor-specific biological relevance. The pooled HR for OS was 2.54, indicating a ~ 2.5-fold higher risk of death for patients with high IINS. These associations were generally consistent across subgroups and robust in sensitivity analyses, though some variability existed due to cancer type and treatment methods. Despite demonstrating prognostic relevance, the prognostic value of IINS should be interpreted cautiously given residual heterogeneity and study limitations.

In hepatocellular carcinoma, tumor development typically occurs in the context of chronic liver disease, where persistent inflammation, immune dysregulation, and progressive malnutrition coexist. Chronic viral hepatitis, cirrhosis, and fibrosis create a pro-tumorigenic microenvironment characterized by sustained cytokine release, recruitment of immunosuppressive cells, and impaired anti-tumor immunity. In this setting, biomarkers integrating inflammatory burden, immune status, and nutritional reserve such as IINS may more accurately capture the systemic milieu that drives tumor progression, thereby exhibiting stronger prognostic discrimination [28, 29]. Similarly, endometrial cancer, particularly in the context of obesity and metabolic syndrome, is closely linked to chronic low-grade inflammation and immune imbalance [30]. Adipose tissue-derived cytokines, insulin resistance, and estrogen-driven signaling contribute to a pro-inflammatory and pro-proliferative environment [31]. Moreover, alterations in immune surveillance and nutritional status may further influence tumor behavior [32]. As a result, composite indices like IINS that incorporate multidimensional host factors may be especially sensitive in reflecting disease aggressiveness in this population.

All included studies were conducted in Chinese populations, which may limit the generalizability of IINS across different ethnic and healthcare contexts. Ethnic differences can influence baseline inflammatory and immune parameters, such as hs-CRP levels, lymphocyte counts, and albumin synthesis, due to genetic polymorphisms, dietary habits, and prevalence of chronic comorbidities. Nutritional status and body composition, which directly affect serum albumin and lymphocyte counts, also vary regionally, potentially altering IINS distributions and prognostic thresholds. In addition, differences in healthcare systems, such as accessibility to routine laboratory testing, early cancer detection, and standardized treatment protocols may affect both the components of IINS and patient outcomes. For instance, populations with higher prevalence of obesity, metabolic syndrome, or chronic inflammatory diseases may present systematically higher baseline IINS values, which could influence its predictive performance.

Additionally, although we applied the trim-and-fill method to adjust for publication bias, the significant Egger’s test results suggested that small-study effects may exist. Such effects could arise if smaller studies with non-significant or negative findings remain unpublished, potentially leading to an overestimation of the pooled HRs. While the sensitivity analyses indicated that the pooled estimates were robust to the exclusion of any single study, we acknowledged that the presence of unpublished negative studies could moderately attenuate the observed associations between high IINS and adverse survival outcomes. To alleviate this concern, future meta-analyses should include more studies to further verify the prognostic relationship of IINS in cancers. Nevertheless, the consistency of results across large-scale studies and multivariable analyses in our current dataset supports the overall reliability of the observed prognostic effect of IINS.

IINS embodied a systemic snapshot of the interplay between host metabolism, inflammation, and immune competence, and its association with cancer prognosis is largely mediated by its influence on the tumor immune microenvironment (TME). The TME is not a passive background but an active immunological battlefield where tumor cells, stromal components, and infiltrating immune populations continuously interact. Within this context, the IINS parameters reflect distinct immune-modulatory processes.

IINS reflects the intricate and dynamic interplay among systemic inflammation, immune competence, and nutritional status. Chronic inflammation is increasingly recognized as a hallmark of cancer, driving tumor initiation, proliferation, angiogenesis, invasion, and immune evasion [33]. Elevated hs-CRP, a central component of IINS, serves as a surrogate marker of systemic inflammatory burden [34]. Sustained inflammation can induce T-cell exhaustion through upregulation of inhibitory checkpoints, impair antigen presentation by dendritic cells, and promote the expansion of immunosuppressive populations including regulatory T cells, myeloid-derived suppressor cells, and tumor-associated macrophages, thereby creating a permissive environment for tumor progression and metastasis [35–37]. Lymphocytes, particularly CD8⁺ cytotoxic T cells and natural killer cells, are central effectors of anti-tumor immunity. Lymphopenia, as captured by IINS, reflects impaired immune surveillance, diminished cytotoxic potential, and reduced cytokine secretion such as interferon-gamma, all of which compromise tumor cell clearance [38–40]. Concurrently, hypoalbuminemia not only indicates malnutrition but also reflects a chronic inflammatory and catabolic state, marked by elevated pro-inflammatory cytokines and increased protein degradation, which can further impair immune function and tissue repair [41–45]. Additionally, nutritional deficits influence the bioavailability of essential amino acids, micronutrients, and antioxidants, which are critical for lymphocyte proliferation, antibody production, and the maintenance of redox balance, thereby linking metabolic status directly to anti-tumor immunity [46–50]. Taken together, high IINS reflects a systemic milieu dominated by chronic inflammation, nutritional deficiency, and immunosuppression, all of which converge to shape a TME favorable to tumor growth and resistant to immune-mediated eradication. The mechanistic link not only explains the robust association between IINS and clinical outcomes but also highlights its potential role in predicting responsiveness to emerging immunotherapeutic strategies.

The innovative value of IINS primarily lies in its multidimensional assessment system. By integrating key pathophysiological indicators, such as the patient’s inflammatory response, immune function, and nutritional status, IINS constructs a more comprehensive prognostic model. This approach overcomes the limitations of relying on a single indicator and provides a more accurate reflection of the overall pathophysiological state of cancer patients. By integrating multiple parameters, IINS minimizes the influence of physiological fluctuations or measurement errors that often affect single biomarkers. Clinically, all components of IINS are derived from routine blood tests, requiring neither specialized testing equipment nor costly reagents. The per-test cost is substantially lower than that of advanced methods such as genetic profiling, making IINS particularly suitable for widespread use across healthcare settings of varying resource levels.

Limitations

Despite the robust association observed between high IINS and unfavorable survival outcomes, several limitations should be acknowledged. Firstly, all included studies were conducted in China, which may limit the generalizability of our findings to other ethnic populations. Secondly, the overall sample size was relatively modest, and the heterogeneity in OS analysis suggested that differences in cancer types, treatment methods, and analysis type may have contributed to variability in effect estimates. Thirdly, the retrospective nature of the included studies introduced inherent risks of selection bias and unmeasured confounding. Regarding publication bias, although the Begg’s test was non-significant, the Egger’s test indicated potential small-study effects, raising the possibility that unpublished negative studies may exist. Fourthly, the included studies varied in endpoint definitions (DFS, RFS, PFS), introducing conceptual heterogeneity that could influence pooled estimates. Fifthly, the representation of cancer subtypes was limited, and evidence for certain malignancies remains scarce, preventing definitive conclusions across all tumors. Sixthly, this meta-analysis was not registered on a formal platform, which may reduce transparency and increase the risk of selective reporting bias. Seventhly, IINS was non-specific, influenced by comorbidities such as infections or liver dysfunction, and lacked standardized cut-off values, which complicated cross-study comparisons and clinical implementation. Furthermore, the inclusion of different cancer types introduced substantial biological and clinical heterogeneity, given the differences in tumor biology, treatment modalities, and prognostic factors across malignancies. This may limit the interpretability and generalizability of the pooled results. Finally, due to limitations in the original data, we could not fully explore the dynamic changes of IINS during treatment or its detailed associations with specific clinicopathological parameters.

Future perspectives

In conclusion, we confirmed that high IINS was associated with unfavorable survival outcome in cancers. IINS can serve as an effective prognostic indicator for patients with cancers, especially for hepatocellular carcinoma and endometrial cancer. This finding offered a novel approach for cancer prognosis assessment, particularly in resource-limited medical settings. However, several challenges remained before its widespread clinical implementation. These included the lack of standardized cut-off values, variability in measurement timing, and the limited evidence regarding its dynamic changes during treatment. Future well-designed prospective, multicenter studies with diverse populations were warranted to validate the prognostic utility of IINS and to explore its role in longitudinal monitoring and treatment guidance. Standardizing cut-off values and measurement timing would improve reproducibility, while integrating IINS with clinical, pathological, and molecular factors could refine individualized risk stratification. Additionally, combining IINS with tumor-specific molecular or genomic biomarkers may further enhance prognostic accuracy and support precision therapy. Addressing these challenges was crucial for translating IINS into routine clinical application.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Xun Li, Man Li and Jia Yu contributed to the inception and design of the study. Sihui Zhang, Chenxuan Zhang and Guanqi Zhang contributed equally to the literature search, analysis, and writing of this manuscript. Rongqiang Liu contributed to the study design and supervision. All the authors approved the final version of the manuscript.

Funding

None.

Data availability

The data used in this study were obtained from previously published articles, which are cited within the manuscript. No new datasets were generated. All relevant data supporting the findings were available within the article and the cited references.

Declarations

Ethics approval and consent to participate

We followed PRISMA guidelines when conducting this analysis.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.