Colorectal cancer (CRC) accounts for approximately 1.9 million new diagnoses and 930,000 deaths globally each year, making it the third most prevalent malignancy and the second leading cause of cancer mortality worldwide [1,2]. Despite advances in endoscopic screening programmes, a significant proportion of patients continue to present with locally advanced or metastatic disease, at which point therapeutic options are substantially constrained [3].
Tissue biomarkers offer a complementary approach to staging and risk stratification, potentially enabling personalised surveillance strategies and earlier therapeutic intervention. Carcinoembryonic antigen (CEA) and CA19-9 have demonstrated utility as prognostic serum markers in advanced CRC, yet their IHC-based expression profiles in early-stage disease remain incompletely characterised [4,5]. CDX2, a homeodomain transcription factor critical to intestinal epithelial differentiation, has emerged as a promising independent predictor of disease recurrence across colorectal cancer stages [6].
Microsatellite instability (MSI) status has gained particular clinical relevance since the approval of pembrolizumab for MSI-High (MSI-H) solid tumours, yet its interaction with conventional IHC biomarkers in early-stage CRC cohorts from diverse geographical populations has not been systematically studied [7,8]. This multi-centre cohort study was therefore designed to address this evidence gap by characterising a curated six-biomarker IHC panel across geographically and ethnically diverse early-stage CRC specimens.
Study Design and Setting. This was a retrospective multi-centre cohort study conducted across three academic tertiary referral centres: Apollo Hospitals Research Division (Chennai, India), Hospital Clínic de Barcelona (Barcelona, Spain), and Lagos University Teaching Hospital (Lagos, Nigeria). Ethical approval was obtained from each institutional review board (Ref: AH-2023-CRC-0041, HCB-2023-0088, LUTH-REC/2023/007), and all patients provided written informed consent for tissue use in research.
Patient Selection. Patients aged ≥18 years with histologically confirmed Stage I or Stage II colorectal adenocarcinoma undergoing curative-intent surgical resection between January 2023 and December 2024 were eligible. Exclusion criteria included prior colorectal malignancy, receipt of neoadjuvant chemotherapy or radiotherapy, hereditary CRC syndromes (FAP, Lynch syndrome), and unavailability of archived tissue blocks of adequate quality.
Immunohistochemical Analysis. Formalin-fixed, paraffin-embedded (FFPE) tissue microarray (TMA) blocks were constructed from representative tumour cores (2 mm diameter, triplicate sampling) and matched normal mucosa. IHC staining was performed using validated commercially sourced primary antibodies: anti-CEA (clone COL-1, Leica Biosystems), anti-CA19-9 (clone 1116-NS-19-9, Novus Biologicals), anti-KRAS (clone EP173Y, Abcam), anti-CDX2 (clone EPR2764Y, Abcam), and anti-MUC2 (clone CCP58, DAKO). MSI status was assessed by PCR-based fragment analysis and confirmed by IHC for MLH1, MSH2, MSH6, and PMS2 mismatch repair proteins.
A total of 284 patients were enrolled (mean age 61.4 ± 8.7 years; 52.1% male). Tumour location was predominantly left-sided (58.8%), with 22.9% caecal/right-sided and 18.3% transverse/splenic flexure. Stage I and Stage II proportions were 41.5% and 58.5% respectively. Median follow-up was 22.4 months (IQR 16.8–28.1 months).
| Characteristic | Stage I (n=118) | Stage II (n=166) | p-value |
|---|---|---|---|
| Age, mean ± SD (years) | 59.8 ± 9.1 | 62.6 ± 8.3 | 0.012 |
| Male sex, n (%) | 60 (50.8) | 88 (53.0) | 0.714 |
| Left-sided tumour, n (%) | 72 (61.0) | 95 (57.2) | 0.518 |
| CEA overexpression, n (%) | 38 (32.2) | 94 (56.6) | <0.001 |
| CDX2 loss, n (%) | 29 (24.6) | 81 (48.8) | <0.001 |
| MSI-High status, n (%) | 28 (23.7) | 24 (14.5) | 0.048 |
| KRAS mutation, n (%) | 45 (38.1) | 67 (40.4) | 0.693 |
On multivariate logistic regression, CEA overexpression (OR 4.21, 95% CI 2.87–6.18, p <0.001) and CDX2 loss (OR 3.74, 95% CI 2.11–6.63, p <0.001) were independently associated with Stage II disease, after adjusting for age, sex, tumour location, and KRAS mutation status. MSI-H status was inversely associated with stage progression (OR 0.41, 95% CI 0.22–0.77, p = 0.006).
Our findings corroborate and extend prior literature demonstrating that IHC-based CEA overexpression represents a robust marker of disease aggressiveness in CRC, even within the early-stage disease stratum [9]. The particularly strong association observed between CDX2 loss and stage progression across all three geographically distinct cohorts suggests that this biomarker may reflect a more fundamental disruption in intestinal epithelial homeostasis that transcends ethnic and environmental risk factor heterogeneity [10].
The improved survival outcomes in MSI-H tumours observed in our cohort are consistent with the established immunogenic phenotype of MSI-H CRC and support the prognostic utility of MSI testing even in early-stage disease, where its implications for surveillance intensity and adjuvant chemotherapy decisions are increasingly recognised [11,12]. The combined biomarker panel AUC of 0.89 compares favourably with previously published single-marker panels and supports the concept that multiparametric IHC profiling may augment conventional pathological staging in risk-stratifying early-stage CRC patients.
A combined IHC biomarker panel comprising CEA overexpression, CDX2 loss, and MSI status demonstrates strong discriminatory power for early-stage CRC progression risk stratification in a geographically diverse multi-centre cohort. These findings support the prospective validation of this panel in larger studies as a complement to current pathological staging systems, with the potential to guide personalised surveillance and adjuvant treatment decisions in Stage I–II colorectal adenocarcinoma.
This work was supported by the Interdisciplinary Cancer Research Grant (ICRG-2022-044) and the European Oncology Research Collaborative Fund (EORC/2022/CRC/19). The authors declare no conflicts of interest.
| Variable | Odds Ratio | 95% CI | p-value |
|---|---|---|---|
| CEA overexpression | 4.21 | 2.87 – 6.18 | <0.001 |
| CDX2 loss | 3.74 | 2.11 – 6.63 | <0.001 |
| MSI-High status | 0.41 | 0.22 – 0.77 | 0.006 |
| KRAS mutation | 1.18 | 0.74 – 1.88 | 0.489 |
| Age (per decade) | 1.24 | 0.98 – 1.57 | 0.072 |
| Male sex | 1.09 | 0.71 – 1.67 | 0.699 |
| Panel | AUC | 95% CI | Sensitivity | Specificity |
|---|---|---|---|---|
| CEA alone | 0.74 | 0.68–0.80 | 68.4% | 72.1% |
| CDX2 alone | 0.71 | 0.65–0.77 | 64.2% | 70.3% |
| MSI status alone | 0.63 | 0.56–0.70 | 55.8% | 68.6% |
| Combined panel (CEA + CDX2 + MSI) | 0.89 | 0.85–0.93 | 83.1% | 84.7% |