Diagnostic efficacy of pan-immune inflammation value for ulcerative colitis activity: a study based on multi-indicator comparison and diagnostic grey zone validation

Abstract

Background

Faecal calprotectin (FC) has limitations including inconvenient sampling and limited accessibility, while several blood markers derived from complete blood counts demonstrate suboptimal diagnostic performance in monitoring ulcerative colitis (UC) disease activity. There is an urgent need for superior noninvasive biomarkers. The pan-immune inflammation value (PIV), a novel composite index derived from complete blood counts, has shown prognostic value in malignancies but remains unexplored in UC activity assessment.

Methods

A retrospective analysis was conducted on 196 patients with UC (145 with active disease, 51 in remission), with 136 patients forming the FC subcohort. The PIV, systemic immune inflammation index (SII), systemic inflammation response index (SIRI), neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and FC were compared using receiver operating characteristic curves. Sensitivity analyses and a diagnostic grey zone (defined as ± 15% around the optimal NLR and PLR cut-offs) were employed to assess robustness.

Results

Patients with active UC demonstrated higher PIV, SII, SIRI, NLR, and PLR compared with those in remission. The area under the curve (AUC) for PIV was 0.731, outperforming other CBC-derived blood markers and showing numerically comparable performance with FC (AUC = 0.764, 95% confidence interval: 0.668–0.860), with no statistically significant difference between the two (P = 0.547). Within the diagnostic grey zone (± 15%) where NLR and PLR were inconclusive, PIV was the only haematological marker retaining statistical significance in the full cohort, albeit with only fair discriminatory ability (AUC = 0.664, 95% CI: 0.539–0.789; P = 0.010). The primary diagnostic performance findings remained robust after winsorization, propensity score matching, and bootstrap resampling validation.

Conclusion

PIV represents a promising low-cost, blood-derived biomarker for assessing UC activity. In clinical settings where FC is unavailable or stool sampling is not feasible, PIV may provide supplementary discriminative information. However, its role as a substitute for FC or as a tool to guide decision-making in diagnostically uncertain scenarios requires confirmation in prospective, externally validated studies before any clinical application.

Trial registration

Not applicable. This study is a retrospective observational study and does not report the results of a health care intervention.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12876-026-05082-z.

Introduction

Ulcerative colitis (UC) is a chronic, recurrent inflammatory bowel disease of unknown aetiology, primarily characterised by continuous inflammatory damage to the colonic and rectal mucosa. With the advancement of the “treat-to-target” concept, the treatment goal for UC has evolved from mere symptom control to achieving deeper mucosal healing [1]. Against this backdrop, accurately assessing disease activity has become a critical prerequisite for developing personalised treatment plans, determining prognosis, and adjusting therapeutic strategies. Assessment bias may lead to undertreatment or overtreatment, significantly increasing the risk of recurrence and the burden on healthcare resources. Therefore, there is an urgent clinical need for an assessment tool that is both accurate and accessible.

Currently, endoscopy combined with tissue biopsy remains the gold standard for assessing UC disease activity. However, its invasiveness, high cost, and operational complexity limit its use as a routine monitoring tool [2]. Traditional blood markers such as erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), while convenient to test, have limited sensitivity and specificity, making it difficult to accurately reflect local intestinal inflammation [3]. Faecal calprotectin (FC), as an intestine-specific biomarker, shows significant correlation with both clinical and endoscopic activity in UC. It has demonstrated diagnostic value for assessing disease activity and predicting mucosal healing [4]. However, its clinical adoption remains limited owing to suboptimal patient compliance with stool collection, relatively high assay costs, and prolonged turnaround times.

Given these limitations, novel systemic immune inflammation markers derived from complete blood counts have garnered increasing clinical attention, including the systemic immune inflammation index (SII), systemic inflammation response index (SIRI), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR). These markers demonstrate significant potential in clinical applications due to their affordability, accessibility, and reproducibility [5–8]. Among them, the pan-immune inflammation value (PIV) theoretically provides a more comprehensive reflection of systemic immune inflammation and coagulation status by integrating four parameters: neutrophils, monocytes, platelets, and lymphocytes. PIV was initially validated as an independent prognostic predictor in colorectal cancer. Fucà et al. demonstrated that elevated PIV levels were significantly associated with worse overall survival and progression-free survival in colon cancer patients [9]. Subsequently, its prognostic value has been corroborated across multiple diseases. In coronary heart disease patients, PIV was independently associated with the risk of major adverse cardiovascular events [10]. In solid tumours such as pancreatic and lung cancers, PIV has also been confirmed as a reliable prognostic biomarker [11, 12]. These studies provide a biologically plausible basis for PIV as a comprehensive indicator of systemic immune-inflammatory status, suggesting its potential clinical utility in other inflammation-related diseases. Specifically in UC, a recent study [13] reported that PIV effectively discriminates UC patients from healthy controls with strong diagnostic performance (AUC = 0.812) and serves as an independent predictor of UC diagnosis. That study examined the relationship between PIV and endoscopic scoring systems (UCEIS and DUBLIN), but it was primarily designed to distinguish UC patients from healthy individuals rather than to systematically evaluate PIV for monitoring disease activity. Furthermore, no studies have directly compared the diagnostic performance of PIV with that of FC in assessing UC activity, nor has any research explored whether PIV provides incremental value in the diagnostic grey zone where traditional markers NLR and PLR show limited discriminatory ability. Addressing this gap may help inform noninvasive monitoring strategies for UC and for clinical scenarios with limited resources or ambiguous traditional biomarker results.

This study aimed to systematically investigate the diagnostic efficacy and clinical utility of PIV in assessing UC disease activity. Specific objectives included: (1) to compare the diagnostic efficacy of PIV with other blood inflammatory indices (SII, SIRI, NLR, PLR) across the full cohort to evaluate whether PIV demonstrates superior diagnostic performance; (2) to validate PIV’s reliability using FC as a reference in a subcohort with FC data, exploring its potential to replace FC for rapid screening; (3) to explore PIV’s diagnostic utility in the “diagnostic grey zone” where NLR and PLR yield inconclusive results, supported by sensitivity analyses, aiming to provide new scientific evidence for disease assessment in clinically uncertain scenarios.

Materials and methods

This observational study was designed, conducted, and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [14].

Study design and research population

This retrospective study included 196 patients diagnosed with UC at the Affiliated Hospital of Nanjing University of Chinese Medicine between December 2022 and October 2025. Disease activity was graded using the Mayo Endoscopy Score (MES): patients with MES 0–1 were assigned to the remission group (n = 51), while those with MES ≥ 2 were assigned to the active group (n = 145). A subcohort analysis was further conducted on 136 patients with available FC measurement data.

Inclusion Criteria: (1) Diagnosed with UC according to the Chinese Guidelines for the Diagnosis and Treatment of Ulcerative Colitis (2023·Xi’an) [15]; (2) Complete blood count (CBC) performed within 7 days prior to endoscopy; (3) Complete clinical documentation, including medical records, endoscopy findings, and laboratory test data; (4) Age 18–75 years.

Exclusion Criteria: (1) Other concurrent inflammatory bowel disease (e.g., Crohn’s disease) or non-infectious enteritis (e.g., ischaemic colitis, drug-induced enteritis); (2) Infectious enteritis (e.g., shigellosis, bacterial dysentery, amoebic enteritis/dysentery); (3) History of malignancy, autoimmune disease, or severe organ damage involving the heart, lungs, brain, kidneys, haematopoietic system, or immune system; (4) Severe infection or severe structural gastrointestinal disease; (5) Use of immunosuppressants or glucocorticoids within 1 month prior to enrolment.

Data on 5-aminosalicylate (5-ASA) and biologics use were not systematically available from the retrospective records and were therefore not included in the analysis.

This study protocol was approved by the Ethics Committee of the Affiliated Hospital of Nanjing University of Chinese Medicine (Approval No.: 2023NL-233-02). Given the retrospective observational design of this study, all patient data were fully anonymised prior to analysis, and no identifiable personal information was retained, thereby protecting patients’ rights and privacy. Due to the retrospective nature of this study, the requirement for written informed consent was formally waived by the Ethics Committee. The study strictly adhered to the relevant provisions of the Declaration of Helsinki.

Data collection and processing

Data collection

Demographic data (age, sex), clinical characteristics (disease duration, smoking history), endoscopic findings, and laboratory test results were extracted from the hospital’s electronic medical record system. Smoking history was categorised into three groups: “never smoker”, “past smoker”, and “current smoker”.

All patients provided venous blood samples in a fasting state. Complete blood counts were performed using a fully automated blood analyser to obtain white blood cell count, neutrophil count, lymphocyte count, monocyte count, and platelet count.

Faecal samples were collected within 3 days of hospital admission. FC levels were measured using a commercially available kit via enzyme-linked immunosorbent assay according to the manufacturer’s instructions. FC detection was performed using multiple commercial assay platforms throughout the study period, resulting in analytical heterogeneity that may have affected the comparative results between PIV and FC. As this study was retrospective, detailed information regarding the kit manufacturer and product number could not be fully obtained. The dataset contained multiple discrete upper detection limits (> 600, > 1800, > 2400, and > 2595 µg/g), suggesting measurements were derived from more than one commercial assay platform without cross-platform standardisation. Due to these detection limits, some FC results were not reported as exact numerical values but were qualitatively labelled as < 30, > 600, > 1800, > 2400, or > 2595 µg/g.

CRP and ESR data were retrieved from the electronic medical record system. However, as these tests were not routinely ordered during the study period, paired measurements within the same 7-day pre-endoscopy window were available for only a minority of patients. CRP and ESR were therefore excluded from the primary comparative analysis.

Definition of inflammatory markers

Based on the CBC results, the following five haematological inflammatory markers were calculated. All calculations were performed using raw count data from the same CBC performed within 7 days prior to endoscopy:

PIV = (Neutrophil Count×Monocyte Count×Platelet Count) / Lymphocyte Count.

SII = (Neutrophil Count×Platelet Count) / Lymphocyte Count.

SIRI = (Neutrophil Count×Monocyte Count) / Lymphocyte Count.

NLR = Neutrophil Count / Lymphocyte Count.

PLR = Platelet Count / Lymphocyte Count.

Data processing

Processing of FC data. For FC values exceeding the analytical detection range, the widely adopted detection limit assignment method was employed to achieve quantitative analysis: Values below the lower limit of detection (< 30 µg/g) were imputed as 15 µg/g (half the lower limit of detection). Values above the upper limits of detection (> 600, > 1800, > 2400, and > 2595 µg/g) were imputed as the corresponding upper limits (600, 1800, 2400, and 2595 µg/g), respectively.

To assess robustness of the results, this study additionally established two alternative imputation strategies in sensitivity analysis (detailed imputation rules are provided in Additional file 1). By comparing receiver operating characteristic (ROC) curves and area under the curve (AUC) values across different strategies, the potential impact of imputation schemes on diagnostic performance was evaluated.

Winsorization of haematological indicators. To control the influence of extreme values and ensure consistency in data processing across analyses, all haematological inflammatory markers (PIV, SII, SIRI, NLR, PLR) in the entire cohort (n = 196) underwent winsorization. The first percentile (P1) and 99th percentile (P99) were calculated for each indicator. Values below P1 were replaced with P1, and values above P99 were replaced with P99 (1% winsorization). These winsorization thresholds (see Additional file 2) were uniformly applied across the full cohort and subgroup analyses to maintain consistency in indicator reference standards. To visually validate the appropriateness and effectiveness of winsorization, box plots and density plots were generated for each haematological inflammation indicator before and after winsorization.

Statistical analysis

Continuous variables were first assessed for normality using the Shapiro-Wilk test. Since the majority of variables did not meet normality assumptions, all continuous variables were presented as median (25th percentile, 75th percentile) [M (Q1, Q3)]. Intergroup comparisons were performed using the Mann-Whitney U test. Categorical variables were expressed as frequencies (percentages) [n (%)], and intergroup comparisons were performed using the chi-square (χ²) test. For 2 × 2 contingency tables, Fisher’s exact test was used when any cell had an expected count less than 5. For contingency tables larger than 2 × 2, the Fisher-Freeman-Halton exact test was used when more than 20% of cells had an expected count less than 5.

ROC curves were constructed to evaluate the diagnostic performance of each inflammatory marker in distinguishing active versus remission groups of UC. The AUC and 95% confidence interval (95% CI) were calculated for each marker. Optimal cut-off values were determined based on the maximum Youden index principle (formula: sensitivity + specificity − 1), with corresponding sensitivity and specificity reported accordingly. Within the FC subcohort, the DeLong test was used to compare AUCs between haematological markers and FC.

To assess the robustness of ROC-derived AUC estimates and optimal cut-off values under potential class imbalance, non-parametric bootstrap resampling with 2,000 replicates was performed, with 95% CIs calculated using the bias-corrected and accelerated (BCa) method. Bootstrap validation with 2,000 resamples was likewise applied to all multivariable logistic regression models to evaluate the stability of estimated odds ratios.

Independent predictors of UC activity were identified using multivariable logistic regression analysis. Given that PIV, SII, SIRI, NLR, and PLR are all mathematically derived from the same blood cell parameters, their simultaneous inclusion in a single model introduces severe intrinsic multicollinearity (variance inflation factors [VIF] > 10 in joint modelling). Therefore, five separate logistic regression models were constructed, each incorporating one inflammatory marker while adjusting for age, sex (forced inclusion), and covariates with P < 0.1 in univariate analysis (including smoking history). All models underwent multicollinearity diagnostics after model fitting to confirm VIF < 5 for all included variables. The independent predictive value of each marker was evaluated through adjusted odds ratios (OR), Wald χ² statistics, and AUC.

Sensitivity analysis was conducted using propensity score matching (PSM) to validate the robustness of the results. Patients in the active and remission groups were matched 1:1 using nearest-neighbour matching without replacement (caliper width set to 0.2) based on age, sex, disease duration, and smoking history. Covariate balance before and after matching was assessed using the standardised mean difference (SMD). |SMD| < 0.1 was considered the ideal threshold, and |SMD| < 0.2 was considered acceptable, per Austin’s recommendation [16].

To evaluate the diagnostic performance of novel inflammatory markers in clinical scenarios where NLR and PLR yield inconclusive results, this study operationalised a “diagnostic grey zone”. This zone refers to a symmetrical interval surrounding the optimal cut-off values of NLR and PLR. In this interval, classification confidence is considered insufficient to guide clinical decision-making [17, 18]. The boundary width was set at ± 15% of each cut-off value, based on two considerations. First, from a theoretical standpoint, no established grey-zone framework has been proposed or validated for NLR or PLR in UC disease activity assessment, and no consensus threshold width has been reported in the literature. The ± 15% width was therefore informed by the clinical rationale that the interval must be sufficiently narrow to avoid missing true positives while sufficiently wide to capture genuine diagnostic uncertainty. Second, to ensure adequate statistical power within the grey zone, this study evaluated the number of eligible patients across candidate thresholds in the full cohort: 60 patients fell within ± 10%, 81 within ± 15%, and 110 within ± 20%. The ± 15% threshold was selected as a pragmatic primary choice that balances statistical adequacy with conceptual parsimony. The ± 15% threshold should be regarded as exploratory. Its clinical validity and generalisability are discussed further in the Limitations section.

To assess the robustness of the grey-zone definition, sensitivity analyses were conducted by redefining the zone using stricter (± 10%) and broader (± 20%) thresholds, and the diagnostic performance of PIV and other markers was reassessed within each subgroup. Consistency of results across thresholds was evaluated to determine whether PIV’s diagnostic value within the grey zone is sensitive to the specific threshold selection.

Statistical analyses were performed using IBM SPSS Statistics software (version 31.0.1) and R software (version 4.4.1).

All statistical tests were two-tailed, and P values < 0.05 were considered statistically significant.

Results

Baseline characteristics of the study population

A total of 196 patients with UC were enrolled as the full cohort, of whom 51 were in remission and 145 had active disease. Among these, 136 patients with available FC measurements formed the FC subcohort, including 33 in remission and 103 with active disease. Baseline characteristics of both cohorts are summarised in (Table 1).

Table 1.

Baseline characteristics of the study population

Characteristics	Full Cohort (n = 196)			FC subcohort (n = 136)
	Remission (n = 51)	Active (n = 145)	P value	Remission (n = 33)	Active (n = 103)	P value
Age (years)	51.0 (41.0, 64.0)	45.0 (35.0, 59.0)	0.060	50.0 (38.0, 64.5)	44.0 (35.0, 57.0)	0.121
Sex, n(%)			0.392			0.427
Female	19 (37.3%)	64 (44.1%)		12 (36.4%)	46 (44.7%)
Male	32 (62.7%)	81 (55.9%)		21 (63.6%)	57 (55.3%)
Disease duration (years)	7.0 (4.0, 12.0)	6.0 (3.0, 9.0)	0.182	7.0 (4.0, 12.0)	6.0 (3.0, 10.0)	0.443
Smoking history, n(%)			0.049			0.123
Never smoker	45 (88.2%)	135 (93.1%)		30 (90.9%)	94 (91.3%)
Past smoker	5 (9.8%)	3 (2.1%)		3 (9.1%)	3 (2.9%)
Current smoker	1 (2.0%)	7 (4.8%)		0 (0.0%)	6 (5.8%)
PIV	112.2 (76.7, 227.2)	233.8 (138.2, 477.5)	< 0.001	109.0 (75.6, 245.9)	243.5 (145.9, 521.9)	< 0.001
SII	378.4 (256.6, 565.4)	588.3 (351.7, 883.9)	< 0.001	378.4 (257.4, 565.2)	609.9 (355.2, 994.7)	< 0.001
SIRI	0.61 (0.42, 0.93)	1.04 (0.60, 1.69)	< 0.001	0.62 (0.42, 1.01)	1.12 (0.63, 1.80)	< 0.001
NLR	1.76 (1.36, 2.68)	2.19 (1.53, 3.25)	0.023	1.76 (1.36, 2.70)	2.39 (1.68, 3.53)	0.023
PLR	132.1 (94.0, 156.5)	141.9 (114.0, 195.5)	0.004	125.3 (98.7, 154.1)	138.8 (113.7, 187.8)	0.043
FC (µg/g)	–	–	–	41.0 (15.0, 124.3)	250.5 (88.8, 593.0)	< 0.001

Data are presented as median (Q1, Q3) for continuous variables and n (%) for categorical variables. Bold P values indicate statistical significance (P < 0.05)

Demographic and clinical characteristics were comparable between patients with active disease and those in remission in both the full cohort and the FC subcohort, with no statistically significant differences in age, sex, or disease duration (all P > 0.05). However, in the full cohort, the proportion of never smokers was higher in the active group than in the remission group (93.1% vs. 88.2%; P = 0.049).

With respect to inflammatory markers, active-phase patients exhibited significantly higher levels of PIV, SII, SIRI, NLR, and PLR than remission patients in both the full cohort and the FC subcohort (all P < 0.05), with the most pronounced differences observed for PIV, SII, and SIRI (all P < 0.001). In the FC subcohort, FC levels were also significantly elevated in active-phase patients compared with those in remission, with median values of 250.5 µg/g (interquartile range [IQR]: 88.8–593.0) versus 41.0 µg/g (IQR: 15.0–124.3), respectively (P < 0.001).

Diagnostic performance analysis

Diagnostic performance of each inflammatory marker across the full cohort

In the full cohort of 196 patients, ROC analysis was performed to evaluate the diagnostic accuracy of PIV, SII, SIRI, NLR, and PLR in discriminating active UC from remission (Fig. 1; Table 2). All five haematological inflammatory markers demonstrated statistically significant diagnostic value for active UC (AUC > 0.5 and P < 0.05 for all). Among them, PIV achieved the highest AUC of 0.731 (95% CI: 0.655–0.808), followed by SII (0.691, 95% CI: 0.611–0.772), SIRI (0.689, 95% CI: 0.610–0.768), PLR (0.634, 95% CI: 0.550–0.719), and NLR (0.607, 95% CI: 0.520–0.693).

Fig. 1.

ROC curves for diagnostic performance. A ROC curves of haematological inflammatory indicators in the full cohort (n = 196) B Comparison of ROC curves between haematological indicators and FC in the FC subcohort (n = 136). Pairwise DeLong tests were performed to compare differences in AUC values

Table 2.

Diagnostic performance of inflammatory indicators for predicting active UC

Variable	AUC	Cut-off	Sensitivity(%)	Specificity(%)	Youden	95% CI
Full Cohort
PIV	0.731	121.990	80.0	58.8	0.388	0.655–0.808
SII	0.691	603.485	49.0	82.4	0.313	0.611–0.772
SIRI	0.689	1.042	49.7	82.4	0.320	0.610–0.768
NLR	0.607	1.802	65.5	54.9	0.204	0.520–0.693
PLR	0.634	166.274	34.5	90.2	0.247	0.550–0.719
FC subcohort
FC(µg/g)	0.764	127.800	68.0	78.8	0.467	0.668–0.860
PIV	0.725	121.775	82.5	54.5	0.371	0.628–0.822
SII	0.698	594.407	52.4	81.8	0.342	0.599–0.798
SIRI	0.699	1.088	51.5	81.8	0.333	0.603–0.796
NLR	0.631	1.772	72.8	54.5	0.274	0.525–0.738
PLR	0.617	193.240	24.3	97.0	0.212	0.513–0.722

 The 95% CIs of all AUCs did not include 0.5, indicating statistical significance (P < 0.05). Cut-off values represent the optimal threshold determined by the maximum Youden index

Bootstrap validation with 2,000 resamples confirmed the stability of these AUC estimates, with bootstrap-derived median AUCs differing from original values by ≤ 0.003 (Additional file 3). Pairwise DeLong test comparisons revealed that PIV had a significantly higher AUC than all other markers (P < 0.05 for all comparisons; Additional file 4). SII and SIRI showed comparable AUC values (P = 0.920), whereas both were significantly higher than that of NLR (SII vs. NLR: P < 0.001; SIRI vs. NLR: P = 0.003).

Regarding optimal cut-off values, PIV at a threshold of 121.99 yielded a sensitivity of 80.0% and a specificity of 58.8%. SII and SIRI, at their respective optimal cut-offs, both achieved a specificity of 82.4% with sensitivities of 49.0% and 49.7%, respectively. NLR at a cut-off of 1.80 yielded a sensitivity of 65.5% and a specificity of 54.9%, while PLR at a cut-off of 166.27 yielded a sensitivity of 34.5% and a specificity of 90.2%.

Diagnostic performance of various indicators in the FC subcohort

In the FC subcohort, ROC curve analysis results (Table 2) showed that FC demonstrated the highest diagnostic performance (AUC = 0.764, 95% CI: 0.668–0.860). The DeLong test showed no statistically significant difference between PIV (AUC = 0.725, 95% CI: 0.628–0.822) and FC (P = 0.547), and pairwise comparisons between FC and the remaining haematological indicators likewise yielded no significant differences (all P > 0.05; Additional file 5). These comparison results may be affected by the analytical heterogeneity of FC from multiple commercial assay platforms, and the non-significant difference should not be interpreted as evidence of diagnostic equivalence between PIV and FC.

Among haematological indicators, PIV maintained optimal diagnostic performance, with AUC significantly higher than that of NLR (AUC = 0.631, P < 0.05) and that of PLR (AUC = 0.617, P < 0.05). Compared with SII (AUC = 0.698) and SIRI (AUC = 0.699), PIV showed a numerically higher AUC, though the differences were not statistically significant (P > 0.05).

At the optimal cut-off value of 121.78, PIV demonstrated a sensitivity of 82.5% and a specificity of 54.5%, whereas FC exhibited a sensitivity of 68.0% and a specificity of 78.8%.

ROC curves for the full cohort and the FC subcohort are shown in (Fig. 1).

Diagnostic efficacy of each indicator in the diagnostic grey zone subgroup

Full Cohort Analysis. Within the full cohort, 81 patients (41.3%) fell within the ± 15% diagnostic grey zone. In this subgroup, PIV was the only haematological indicator retaining statistically significant diagnostic value (AUC = 0.664, 95% CI: 0.539–0.789, P = 0.010), whereas SII and SIRI did not achieve statistical significance (SII: P = 0.059; SIRI: P = 0.075) (Table 3). When validated using alternative grey zone definitions (± 10% and ± 20%), PIV yielded higher AUC values than SII and SIRI across all definitions. Within the ± 10% subgroup (n = 60), PIV achieved an AUC of 0.699 (P = 0.008).

Table 3.

Diagnostic performance of inflammatory indicators in the ± 15% diagnostic grey zone subgroup

Variable	n (%)	AUC (95% CI)	P-value
Full Cohort (Grey Zone ± 15%)
PIV	81 (41.3%)	0.664 (0.539–0.789)	0.010
SII		0.626 (0.495–0.756)	0.059
SIRI		0.619 (0.488–0.750)	0.075
FC subcohort (Grey Zone ± 15%)
FC(µg/g)	48 (35.3%)	0.800 (0.645–0.956)	< 0.001
PIV		0.637 (0.459–0.814)	0.132
SII		0.603 (0.424–0.782)	0.260
SIRI		0.607 (0.429–0.785)	0.239

NLR and PLR were used as stratification criteria to define the diagnostic grey zone and were therefore excluded from the comparative performance analysis in this subgroup

FC Subcohort Analysis. In the FC subcohort, FC demonstrated statistically significant diagnostic performance across all three grey zone definitions (± 10%, ± 15%, and ± 20%) (AUC range: 0.769–0.826, all P < 0.01). Among haematological indicators, PIV reached statistical significance only in the ± 20% subgroup (n = 65) (AUC = 0.693, 95% CI: 0.545–0.841, P = 0.010). In the ± 15% (n = 48) and ± 10% (n = 34) subgroups, none of the haematological indicators—PIV, SII, or SIRI—reached statistical significance (all P > 0.05). Specifically, in the pre-specified ± 15% grey zone, PIV showed a numerically higher AUC but no statistically significant discriminatory ability (AUC = 0.637, 95% CI: 0.459–0.814, P = 0.132), which differed from the significant finding in the full cohort. AUC values for all indicators are provided in Table 3. ROC curves are presented in Fig. 2 and Additional file 6.

Fig. 2.

ROC curves of inflammatory indicators in the ± 15% diagnostic grey zone. A Full Cohort (n = 81). The red line represents PIV (AUC = 0.664, P = 0.010). B FC subcohort (n = 48). The blue line represents FC (AUC = 0.800). PIV (AUC = 0.637) is shown for reference

Robustness verification results

Multivariable Logistic Regression Analysis. Given that PIV, SII, SIRI, NLR, and PLR all derive from the same haematological parameters, severe intrinsic multicollinearity exists (VIF > 10 when combined in modelling). This study constructed five independent multivariable logistic regression models to separately validate the independent predictive value of each indicator. After adjusting for age, sex (forced inclusion), and smoking history (univariate P < 0.1), all models demonstrated good stability (VIF < 5). In each model, the corresponding inflammatory marker remained a statistically significant independent predictor of active UC after covariate adjustment. PIV showed an adjusted OR of 1.003 (95% CI: 1.001–1.005, Wald P = 0.004), and SII, SIRI, NLR, and PLR showed similarly significant associations within their respective models (all Wald P < 0.05, Table 4). Because each model contains a different inflammatory marker, the model-level AUCs in Table 4 are not directly comparable as measures of relative marker performance and are reported for descriptive purposes only. Direct between-marker comparisons are based on Table 2.

Table 4.

Independent predictive value of inflammatory indices for active UC

Model	Variable	Adjusted OR (95%CI)	Wald P	VIF	AUC
Model 1	PIV	1.003 (1.001–1.005)	0.004	1.024	0.746 (0.669–0.823)
Model 2	SII	1.002 (1.001–1.003)	0.001	1.004	0.727 (0.647–0.807)
Model 3	SIRI	2.239 (1.274–3.935)	0.005	1.031	0.724 (0.646–0.803)
Model 4	NLR	1.391 (1.033–1.873)	0.030	1.022	0.672 (0.585–0.759)
Model 5	PLR	1.011 (1.004–1.019)	0.002	1.016	0.681 (0.598–0.764)

Each multivariable model was adjusted for age, sex, and smoking history. VIF < 5 indicates no multicollinearity within each independent model. Model-level AUCs are reported for descriptive purposes only and are not directly comparable across models

PSM Analysis. To control for potential confounding effects of age, sex, disease duration, and smoking history, a 1:1 PSM was performed in this study, ultimately generating a matched cohort comprising 98 patients (49 in the active group and 49 in the remission group). Post-matching covariate balance was acceptable for all four variables (all |SMD| < 0.2). Age, sex, and smoking history additionally met the stricter |SMD| < 0.1 threshold, while disease duration showed a residual |SMD| of 0.179. Although this exceeded the ideal threshold of 0.1, it remained below the conventional boundary of 0.2 and was therefore considered acceptable for subsequent analyses. Residual imbalance on this variable, however, cannot be fully excluded (Fig. 3; Additional file 7).

Fig. 3.

Covariate balance before and after PSM. Note: Blue circles (Adjusted) and red triangles (Unadjusted) represent SMDs before and after 1:1 nearest-neighbor PSM (caliper = 0.2). The vertical dashed line at 0.1 indicates the ideal balance threshold (|SMD| < 0.1). The "distance" row represents the propensity score and is not subject to the balance criterion

In ROC analysis of the matched cohort, PIV demonstrated statistically significant diagnostic value (AUC = 0.716, 95% CI: 0.613–0.818, P < 0.001). In contrast, traditional inflammatory markers such as PLR (P = 0.063) and NLR (P = 0.097) failed to achieve statistical significance after controlling for confounding factors. Although SII (AUC = 0.682, P = 0.002) and SIRI (AUC = 0.673, P = 0.003) remained statistically significant, PIV numerically demonstrated the highest AUC value (Table 5).

Table 5.

ROC analysis of haematological indices in the PSM cohort (n = 98)

Variable	P-value	AUC	95% CI
PIV	< 0.001	0.716	0.613–0.818
SII	0.002	0.682	0.576–0.788
SIRI	0.003	0.673	0.566–0.780
PLR	0.063	0.609	0.497–0.721
NLR	0.097	0.597	0.483–0.712

The P-values indicate whether the AUC is significantly different from 0.5. Bold values indicate statistical significance (P < 0.05)

Bootstrap internal validation

Bootstrap resampling (2,000 replicates) confirmed the robustness of all primary estimates. Median bootstrap AUCs differed from original values by ≤ 0.003 across all markers in both cohorts (Additional file 3). Bootstrap-corrected ORs from multivariable logistic regression deviated by less than 1% from original estimates for all model variables, including PIV, age, sex, and duration (Additional file 8).

Sensitivity analysis and data robustness verification

First, to assess the impact of FC data imputation strategies on outcomes, this study compared the original strategy with two alternative approaches (specific strategy details are provided in Additional file 1). Analysis revealed that the diagnostic performance of FC remained highly consistent across different assignment strategies (AUC = 0.764, 95% CI: 0.668–0.860), with ROC curves showing near-complete overlap. The diagnostic performance of PIV was also unaffected by FC data processing methods (AUC range: 0.723–0.725) (Fig. 4A).

Fig. 4.

Sensitivity analyses validating the robustness of data processing strategies. A ROC curves of FC under three imputation strategies for out-of-range values. Near-complete overlap of the three curves indicates stable diagnostic performance across imputation methods. B Distribution of five inflammatory indices (PIV, SII, SIRI, NLR, PLR) before and after 1% winsorization. (B-i) Density plots show that winsorization primarily truncated the right tails without altering the core distribution morphology. (B-ii) Box plots show the effective removal of extreme outliers present in the raw data

Second, to assess the influence of extreme outliers, all haematological indicators underwent 1% tail trimming (winsorization at the 1st and 99th percentiles). The AUC of PIV after trimming (AUC = 0.730) was virtually unchanged from the pre-trimming value (AUC = 0.731; Fig. 4B). Visual analysis (Fig. 4B) demonstrated that tail trimming effectively truncated extreme outliers, resulting in a more compact data distribution while largely preserving the distributional differences between the remission and active groups. The diagnostic performance of PIV after trimming (AUC = 0.730) remained virtually unchanged compared to before trimming (AUC = 0.731).

Discussion

Principal findings

This study systematically evaluated the diagnostic value of PIV for monitoring UC disease activity in a retrospective cohort of 196 patients. Among haematological inflammatory indicators, PIV demonstrated the highest diagnostic accuracy in both the full cohort (AUC = 0.731) and the FC subcohort (AUC = 0.725), outperforming SII, SIRI, NLR, and PLR. Multivariable logistic regression confirmed PIV as a statistically significant independent predictor of active UC after adjusting for age, sex, and smoking history, a finding corroborated by propensity score matching analysis. PIV was the only haematological marker retaining statistically significant discriminatory value in the diagnostic grey zone where NLR and PLR were inconclusive in the full cohort (AUC = 0.664, P = 0.010), although this AUC reflects only fair discrimination and PIV should be interpreted as providing supplementary rather than definitive information in this context. This grey zone discriminatory value was not replicated in the FC reference subcohort, where PIV showed no statistically significant performance in the pre-specified ± 15% grey zone (P = 0.132). These findings suggest that PIV may serve as a potential complement to FC, particularly in scenarios where conventional markers yield diagnostically uncertain results and FC testing is unavailable. However, all findings require prospective external validation before clinical application.

Diagnostic performance of PIV relative to haematological comparators

The search for highly effective noninvasive biomarkers has remained a focal point in UC clinical management [19]. In recent years, indicators derived from complete blood counts have been extensively studied due to their accessibility, evolving from traditional single ratios (e.g., NLR and PLR) to composite indices incorporating multiple parameters (e.g., SII and SIRI) [20–23]. Our results partially align with recent findings by Zhang et al. [24] and Cesaro et al. [6], which confirmed significantly elevated levels of these systemic inflammation markers in patients with active disease, further supporting the potential of peripheral blood biomarkers to reflect intestinal mucosal inflammation. However, the diagnostic performance of these markers has yielded inconsistent results across different studies. Compared with the higher predictive values reported by Zhang et al. [24] in a large-scale study (AUCs of 0.812, 0.785, and 0.753 for SII, NLR, and PLR, respectively), the diagnostic performance of corresponding indicators in this study was slightly lower (AUCs of 0.691, 0.607, and 0.634, respectively). This discrepancy may be attributed to sample heterogeneity and the rigorous winsorization of data applied in this study to ensure robust results. Furthermore, while Cesaro et al. [6] found SIRI superior to NLR and PLR, they noted its overall efficacy in the IBD cohort remained “suboptimal” (AUC = 0.627). This may be attributed to the inclusion of Crohn’s disease patients, whose phenotypic variability may have interfered with biomarker specificity. These findings suggest that reliance solely on existing conventional composite indices may fall short of precision monitoring requirements. CRP and ESR, the reference blood-based markers in current UC clinical guidelines, were not included in the comparative analysis for reasons outlined in the Limitations section. The conclusion that PIV demonstrates superior diagnostic performance in this study is therefore limited to the CBC-derived composite indices evaluated and cannot be generalised to other blood-based inflammatory markers.

Among the five haematological markers evaluated in the full cohort (n = 196), PIV demonstrated the highest diagnostic accuracy with an AUC of 0.731. Pairwise DeLong comparisons revealed that PIV significantly outperformed SII (AUC = 0.691, P = 0.045), SIRI (AUC = 0.689, P = 0.012), PLR (AUC = 0.634, P = 0.023), and NLR (AUC = 0.607, P < 0.001). SII and SIRI exhibited comparable diagnostic performance (P = 0.920), both significantly superior to NLR (P < 0.001 and P = 0.003, respectively), while PLR and NLR showed no significant difference (P = 0.513). These exploratory findings suggest that PIV may be a candidate haematological marker for further prospective, multi-centre evaluation in settings where FC testing is inaccessible. A recent study published in Current Medical Research and Opinion first examined the association between PIV and UC, demonstrating strong diagnostic discrimination between UC patients and healthy controls (AUC = 0.812) and identifying PIV as an independent predictor [13]. The present study extends these findings by shifting the clinical question from diagnosis to activity assessment. The lower AUC observed here (0.731) likely reflects the more subtle inflammatory contrast between active and quiescent disease compared with that between diseased and healthy populations. That study found that PIV correlated with UCEIS and DUBLIN scores but not with MES, whereas our MES-based classification still yielded meaningful PIV performance. This discrepancy may be related to the coarse four-tier grading of MES and differences in sample characteristics between the two cohorts.

Compared with SII and SIRI, PIV maintained significant diagnostic value within the grey zone in this study. This advantage may stem from its unique four-parameter composition. Although SII and SIRI also incorporate multiple cell lineages, SII does not include monocytes, whereas SIRI excludes platelets. Consequently, SII may incompletely capture inflammatory amplification driven by monocyte-derived cytokine release, while SIRI may miss platelet-mediated thromboinflammatory signalling. This may contribute to their diminished discriminative performance in diagnostically uncertain scenarios. By simultaneously integrating neutrophils, monocytes, platelets, and lymphocytes, PIV provides a more comprehensive reflection of the overall “immune–inflammatory–coagulation” network. The biological basis for this advantage may lie in the synergistic interplay between monocytes and platelets in amplifying systemic inflammatory responses in UC [25].

From a biological perspective, monocytes, as key innate immune effector cells, not only participate directly in the initiation and perpetuation of mucosal inflammation in UC [26] but also activate systemic inflammatory responses through the secretion of pro-inflammatory cytokines such as IL-6 and TNF-α [27]. Activated platelets, in turn, exacerbate intestinal mucosal injury through the upregulation of CD62P (P-selectin) [28] and further amplify inflammatory signalling through the formation of monocyte–platelet complexes via the CD62P–CD162 pathway [29]. A systematic review and meta-analysis confirmed that platelet counts and activity are significantly altered in UC patients and correlate closely with disease activity [30]. The mathematical structure of PIV enables nonlinear amplification of these monocyte–platelet synergistic signals. When changes in a single cell lineage are subtle and NLR and PLR fall within diagnostically ambiguous ranges, PIV can still sensitively detect underlying inflammatory states through minor synergistic shifts in monocytes and platelets, which may constitute the biological basis for its robust performance in the diagnostic grey zone.

PIV versus FC: complementarity and interpretive cautions

In the FC subcohort, PIV (AUC = 0.725) and FC (AUC = 0.764) demonstrated numerically comparable overall diagnostic efficacy as assessed by the DeLong test (P = 0.547). However, this non-significant result should not be construed as evidence of equivalence. As detailed in the Methods section, FC measurements in this cohort were obtained from multiple commercial assay platforms. This is supported by the discrete upper detection limits of > 600, >1800, > 2400, and > 2595 µg/g. The resulting analytical heterogeneity may have attenuated the apparent AUC gap, thereby contributing to the non-significant comparison.

As a specific marker of local intestinal inflammation, FC directly reflects neutrophil infiltration in the mucosal layer. It has been shown to correlate with endoscopic and histological activity and holds recognised value in predicting mucosal healing [31–33]. In contrast, PIV, as a systemic immune-inflammatory indicator, captures the imbalance within the immune–inflammatory–coagulation network across the entire body. In the FC subcohort, FC exhibited higher specificity (78.8%), aiding in the exclusion of false positives. However, PIV showed superior sensitivity (82.5%), effectively minimising missed active cases. This distinction between local and systemic assessment, and between high specificity and high sensitivity, suggests a potentially complementary clinical role for the two markers. Whether this complementarity translates into additive value in a combined PIV-FC model warrants prospective evaluation.

Diagnostic grey zone: definition, performance, and limitations

This study provided the first operational definition of the diagnostic grey zone for NLR and PLR, defined as ± 15% around the optimal cut-off values, and validated this definition through sensitivity analyses using alternative thresholds (± 10% and ± 20%). Within the ± 15% grey zone of the full cohort (n = 81), PIV was the only haematological marker retaining statistical significance (AUC = 0.664, P = 0.010), whereas SII (P = 0.059) and SIRI (P = 0.075) failed to reach significance. PIV maintained significant diagnostic performance (AUC = 0.699, P = 0.008) even within the most stringent ± 10% subgroup (n = 60), although this AUC indicates only fair discriminatory ability.

However, this significant performance of PIV in the grey zone was not validated in the FC subcohort, the reference standard subgroup of this study. In the pre-specified ± 15% grey zone of the FC subcohort (n = 48), none of the haematological markers, including PIV, reached statistical significance (P = 0.132 for PIV). This discrepancy may be attributed to three key factors: first, the reduced sample size of the FC subcohort grey zone subgroup led to insufficient statistical power to detect moderate effect sizes; second, the analytical heterogeneity of FC assays from multiple commercial platforms may have introduced bias in the reference standard classification; third, the FC subcohort may have inherent clinical heterogeneity compared with the full cohort, as patients with FC testing are more likely to have clinically suspected active disease, narrowing the inflammatory difference between active and remission subgroups.

Nevertheless, the grey zone definition in this study was derived from optimal cut-off values calculated within the same dataset, representing an exploratory approach with inherent internal validation limitations. As no unified standard currently exists for defining diagnostic grey zones, the present methodology provides a reference framework for future research, but further refinement through external cohort validation or predefined clinical misclassification risk thresholds is warranted before clinical application.

Internal validation and robustness of findings

To assess the robustness of our findings against the observed class imbalance (active: remission approximately 3:1), we performed comprehensive bootstrap resampling with 2,000 replicates for both the full cohort and the FC subcohort. The bootstrap-derived median AUCs closely approximated the original estimates for all parameters across both cohorts, with absolute discrepancies ≤ 0.003. Bootstrap median cut-off values showed no systematic bias across most markers. For FC, SIRI, NLR, and PLR, medians were essentially identical to the original thresholds, whereas PIV exhibited a modest upward shift (121.78 to 155.69 in the FC subcohort), consistent with the known sensitivity of the Youden index to class imbalance. Confidence interval widths varied as expected. SIRI and NLR demonstrated relatively narrow intervals, whereas FC, PIV, SII, and PLR showed wider 95% CIs, reflecting uncertainty attributable to the smaller remission sample size. Although the FC cut-off median precisely matched the original value (127.80 µg/g), its 95% CI (41.50–310.35 µg/g) indicates that individual threshold estimates may vary considerably in smaller cohorts. Nonetheless, the stability of the median supports the clinical relevance of 127.8 µg/g as a reasonable diagnostic reference.

Bootstrap validation of the multivariable logistic regression model (2,000 resamples, BCa method) likewise confirmed the stability of parameter estimates. Bootstrap-corrected odds ratios closely approximated the original estimates for all variables, with discrepancies of less than 1% for PIV, age, sex, and disease duration. The bootstrap 95% CI for PIV (1.000–1.008 per unit increase) was broadly consistent with the original interval (1.001–1.005). Although the BCa lower bound approached the null value (OR = 1.000), the bootstrap median OR remained identical to the original estimate (1.003), supporting the stability of PIV as an independent predictor of active disease. In the FC subcohort, bootstrap-derived P-values for pairwise AUC comparisons were highly consistent with the DeLong test results (all P > 0.05), further corroborating the robustness of these findings. Taken together, these bootstrap results support the internal consistency of our primary conclusions. However, internal validation cannot fully substitute for external validation, and future prospective studies with more balanced cohort designs and independent validation cohorts remain warranted.

Limitations and future directions

This study has several limitations. First, the single-centre retrospective design with a limited sample size (n = 196) and an imbalanced active-to-remission ratio (145:51, approximately 3:1) introduces three inter-related sources of potential bias: (i) selection bias inherent to the endoscopy-driven ascertainment process: patients with suspected active disease are more likely to undergo colonoscopy, while those in stable remission are systematically underrepresented, which may inflate the apparent diagnostic performance of the evaluated markers; (ii) residual confounding due to unmeasured variables, including subclinical infections, body mass index, and dietary patterns, cannot be excluded; (iii) the absence of an external validation cohort means that the reported AUC values, optimal cut-offs, and grey-zone thresholds may not generalise to other populations. To assess the robustness of our findings under the observed class imbalance, bootstrap resampling (2,000 replicates) was performed. The bootstrap-derived median AUCs closely approximated the original estimates (absolute discrepancies ≤ 0.003), supporting the stability of the results. Future multi-centre, prospective studies with balanced cohort designs and external validation are warranted.

Second, the diagnostic grey zone definition (optimal cut-off ± 15%) carries two compounding methodological weaknesses. The NLR and PLR cut-offs from which the grey zone is derived were calculated within the same dataset, introducing optimism bias. In addition, the ± 15% window width was chosen on pragmatic sample-size grounds rather than pre-specified clinical misclassification thresholds, with no published UC-specific benchmark available. Furthermore, the significant grey zone performance of PIV was only observed in the full cohort and not replicated in the FC reference subcohort, which further limits the clinical generalisability of this finding. These grey zone findings should therefore be regarded as hypothesis-generating only, and the ± 15% thresholds should not be applied to guide individual patient management without independent prospective validation.

Third, although patients receiving immunosuppressants or glucocorticoids within one month of enrolment were excluded, the use of 5-ASA and biologics was not systematically documented or adjusted for. These agents are known to modulate systemic inflammatory status and peripheral blood cell composition. Drug-stratified analyses were not feasible due to incomplete medication records. Future prospective studies should collect detailed treatment information to enable stratified evaluation of PIV across different therapeutic backgrounds.

Fourth, the use of multiple FC assay platforms across the study period is a key limitation that directly affects the interpretation of our FC-based analyses. This is evidenced by the presence of discrete upper detection limits (> 600, > 1800, >2400, and > 2595 µg/g). This heterogeneity has three specific consequences: (i) the FC reference standard against which PIV was benchmarked is a composite of measurements from non-equivalent analytical platforms, introducing unpredictable bias into the PIV–FC comparison; (ii) the FC optimal cut-off of 127.8 µg/g derived here is not transferable to single-platform clinical settings; and (iii) the non-significant DeLong comparison between PIV and FC (P = 0.547) should not be interpreted as equivalence, because inter-kit analytical variability in the FC arm can artificially narrow the apparent AUC gap. This analytical heterogeneity has been explicitly stated in both the methodology and results sections to avoid misinterpretation of FC as a uniformly standardised reference standard. Future studies should adopt a single, well-characterised FC assay platform with full analytical validation documentation.

Fifth, CRP and ESR were not included in the comparative analysis because their measurements were not routinely recorded across the study period and the proportion of patients with paired CBC–CRP–ESR data obtained within the same pre-endoscopy window was insufficient for a robust comparison. Moreover, the missing data were non-random, as CRP and ESR were preferentially ordered for symptomatic patients. Consequently, the conclusions regarding PIV’s diagnostic performance relative to blood-based markers in this study are restricted to the CBC-derived composite indices evaluated (SII, SIRI, NLR, PLR), and no inference can be drawn regarding PIV’s performance relative to CRP or ESR. Future prospective studies should capture CRP and ESR alongside CBC indices to enable a comprehensive head-to-head comparison.

Sixth, disease activity was classified by MES alone, without paired histological assessment. As the STRIDE-II consensus has established histological healing as a more stringent treatment target beyond endoscopic remission, the diagnostic performance of PIV for identifying histological remission remains unknown. Future prospective studies should incorporate paired endoscopic and histological evaluation to address this clinically relevant question.

The present findings generate several research hypotheses regarding the potential clinical positioning of PIV in UC management, all of which require prospective validation before any clinical application can be considered. PIV may warrant investigation as a candidate monitoring tool in settings where FC is unavailable or stool-sampling compliance is poor, though whether it can adequately substitute for FC in such contexts remains to be established. Within the diagnostic grey zone, PIV retained fair discriminatory ability (AUC = 0.664), suggesting it may carry supplementary information when conventional indices are inconclusive. However, this observation is exploratory and requires external validation. A sequential PIV-then-FC evaluation strategy could be conceptualised as a research hypothesis but remains entirely speculative. We therefore frame these possibilities strictly as research directions and do not recommend changes to current clinical practice on the basis of these retrospective findings.

Conclusion

This study demonstrated that PIV exhibited superior diagnostic performance compared with the CBC-derived composite indices evaluated herein, including SII, SIRI, NLR, and PLR. PIV also demonstrated comparable diagnostic accuracy to FC in the available subcohort (DeLong P = 0.547), though multi-platform assay variability in the retrospective FC data precludes formal equivalence claims. In the diagnostic grey zone, PIV was the only haematological marker to retain statistical significance where NLR and PLR failed in the full cohort; however, this finding was not validated in the FC reference subcohort, where PIV showed no statistically significant discriminatory ability in the pre-specified grey zone. Nevertheless, PIV’s discriminatory performance in this subgroup was only fair (AUC = 0.664), and this grey zone definition remains exploratory given its derivation from internal cut-offs. These findings suggest that PIV is a promising, low-cost, and easily accessible marker for noninvasive UC activity assessment. PIV shows potential as a triage candidate and an FC-sparing alternative in exploratory research settings where FC testing is unavailable. Whether PIV could serve as an alternative to FC, a complementary tool in diagnostically uncertain cases, or a component of composite monitoring strategies remains to be established. Its supplementary value in the diagnostic grey zone requires further validation in prospective studies with standardised FC assays and sufficient statistical power. Prospective multi-centre studies incorporating standardised FC assays, external validation cohorts, and direct comparison with CRP and ESR are needed to confirm PIV’s clinical value and define its role in UC management.

Abbreviations

UC

Ulcerative colitis

FC

Faecal calprotectin

PIV

Pan-immune inflammation value

SII

Systemic immune inflammation index

SIRI

Systemic inflammation response index

NLR

Neutrophil-to-lymphocyte ratio

PLR

Platelet-to-lymphocyte ratio

AUC

Area under the curve

CI

Confidence interval

OR

odds ratio

ESR

Erythrocyte sedimentation rate

CRP

C-reactive protein

ROC

Receiver operating characteristic

VIF

variance inflation factor

PSM

Propensity score matching

SMD

Standardised mean difference

CBC

Complete blood count

5-ASA

5-aminosalicylate

IQR

iIterquartile range

MES

Mayo Endoscopy Score

IBD

iInflammatory bowel disease

BCa

Bias-corrected and accelerated

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

Authors’ contributions

SSX conceived and designed the study, collected and analyzed the data, and drafted the manuscript. MCW contributed to data collection and interpretation of results. AQY contributed to statistical analysis and data visualization. XYM assisted with manuscript revision. JD supervised the study, critically revised the manuscript, and approved the final version. SSX and MCW contributed equally to this work and should be considered co-first authors. All authors read and approved the final manuscript.

Funding

This work was supported by the 2025 Hospital-level Research Project (National Medical Center Construction Special Topic, Grant No. Y25005) and the 2023 Open Project of the National Clinical Research Base of Traditional Chinese Medicine (Jiangsu Provincial Hospital of Chinese Medicine, Grant No. JD2023SZ05). The ongoing project partially covered the publication and experimental costs of this study, and the completed project established its clinical research foundation.

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to patient privacy considerations, but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of the Affiliated Hospital of Nanjing University of Chinese Medicine (2023NL-233-02), and was performed according to the Declaration of Helsinki. The informed consent was waived by the Ethics Committee of the Affiliated Hospital of Nanjing University of Chinese Medicine due to the retrospective design of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.