Glandular epithelium-derived neoplasms account for most malignancies arising within the gastrointestinal (GI) tract. Advances in high-throughput “multi-omics” technologies—most notably through large-scale efforts such as The Cancer Genome Atlas (TCGA)—have transformed our understanding of the molecular landscape of GI cancers.1,2 In particular, colorectal and gastroesophageal adenocarcinomas have been the subject of extensive molecular characterization, now with growing implications for diagnosis, prognosis, and therapy.3,4

While histopathologic assessment remains the cornerstone of cancer classification, molecular profiling has become an essential component of standard patient management. Genomic alterations, microsatellite (MSI) status, and biomarker expression now inform therapeutic strategies in both localized and advanced disease.5

In this review, we aim to bridge the gap between molecular biology and molecular pathology. The former elucidates mechanisms of tumor development and progression, while the latter focuses on how these insights can be applied in daily clinical practice. Although many molecular discoveries expand our understanding of tumor biology, not all are meaningful for clinical implementation currently.4,6 With this in mind, we focus on practical molecular advances with diagnostic, prognostic, or predictive utility in colorectal and gastroesophageal adenocarcinomas. Our goal is to provide a pathologist-centered perspective on relevant biomarkers, molecular classification frameworks, and actionable targets in these common encountered yet molecularly diverse cancers.

Colorectal cancer (CRC) is the third most common cause of cancer-related death in the United States and the fourth leading cause of cancer death worldwide.7, 8, 9 Extensive CRC screening has substantially reduced incidence and mortality through early detection and removal of precancerous adenomas.10 Surgery and chemotherapy are the mainstays in treating CRC; however, long-term survival outcomes have shown modest improvement over time.11 Although histological subtyping remains central to the classification and management of CRC, molecular profiling has increasingly enabled the identification of biologically distinct CRC subgroups with differing prognostic, predictive, and therapeutic implications. Current molecular frameworks primarily divide CRC into hypermutated and non-hypermutated types, with further classification based on key genetic alterations.12,13 A significant advantage of this molecular classification is the identification of biomarkers that guide immunotherapy and targeted treatments, enabling more durable responses and advancing personalized treatment in CRC.

Table 1, Table 2 provide a comprehensive list of CRC subtypes based on their molecular or NCCN guideline-driven biomarker profile14 and the corresponding molecular pathology testing and therapeutic implications.

Approximately 15 % of CRCs exhibit a hypermutated phenotype, defined by a high tumor mutational burden (TMB) and a corresponding increased neoantigen load. These tumors fall into two main molecular categories: microsatellite instability-high (MSI-H) and POLE-mutated.

MSI results from defective DNA mismatch repair (MMR), commonly due to sporadic epigenetic silencing of the MLH1 gene or germline mutations in MLH1, MSH2, MSH6, or PMS2 genes (Lynch syndrome).15 Approximately 13–16 % of sporadic CRCs exhibit MSI, with most showing a CpG island methylator phenotype (CIMP-high). MSI tumors are classified under Consensus Molecular Subtype 1 (CMS1) and, in turn, lead to the accumulation of insertion-deletion loops (IDLs) in repetitive microsatellite sequences, resulting in frameshift mutations in key tumor suppressor genes such as TGFBR2, ACVR2, BAX, and MBD41.16,17 Immunohistochemistry (IHC) for MMR proteins (Fig. 1), PCR-based assays or next-generation sequencing (NGS) have become routine in ascertaining MSI status in CRC patients. MSI-H CRC typically arises in the right colon, demonstrates mucinous or medullary histology and poor differentiation, and carries a dense infiltration of tumor-infiltrating lymphocytes (TILs).18 Paradoxically, these tumors have a favorable prognosis in early-stage disease due to enhanced immunogenicity and cytotoxic T-cell infiltration.18,19 Clinical trials such as KEYNOTE-177 demonstrated that the PD-1 inhibitor antibody pembrolizumab significantly improved progression-free survival over chemotherapy in MSI-H metastatic CRC, establishing it as a frontline therapy for this patient subset.20 Other immune checkpoint inhibitors (ICIs) like nivolumab, alone or combined with ipilimumab, have also demonstrated impressive response durability in second-line settings.21

A smaller subset of CRCs (∼3 %) shows ultra-hypermutated genomes due to exonuclease domain mutations epsilon in POLE or POLD1 and defective proofreading activity in the exonuclease domain of DNA polymerase.2 Unlike MSI-H tumors, POLE-mutant cancers are MSI stable (MSS); however, they show high TMBs (>100 mutations/Mb) due to the accumulation of a high burden of single-nucleotide variants, creating a highly immunogenic neoantigen landscape.22 Burgeoning data demonstrate that POLE-mutated tumors often behave similarly to MSI tumors and report durable responses to PD-1 inhibitors in both localized and metastatic POLE-mutant CRC.23, 24, 25 Additionally, these tumors tend to arise in younger patients and have been reported to display strong MHC-I expression, suggesting a distinct immunobiological footprint.26

The non-hypermutated CRC subtype constitutes approximately 85 % of sporadic CRCs. It demonstrates chromosomal instability (CIN), characterized by aneuploidy, somatic copy number alterations (SCNAs), and structural rearrangements leading to amplifications, deletions, and rearrangements affecting key growth-regulating genes. They include CMS2 (canonical), CMS3 (metabolic), and CMS4 (mesenchymal) subtypes, and several oncogenic signaling cascades dominate this landscape.27

The RAS/RAF/MEK/ERK axis is one of the central pathways mediating CRC oncogenesis through proliferation and survival of the cancerous epithelial cells. Mutations in KRAS (∼40 %) and BRAF (∼10 %) can lead to constitutive activation of downstream signaling through the MAPK and PI3K pathways.28

BRAF mutations occur in 8–10 % of CRCs, with BRAF V600E being the most prevalent mutation. BRAF V600E is a class I mutation that leads to constitutive kinase activity, RAS independence, and monomeric signalling.29 BRAF V600E mutations occur predominantly in right-sided tumors and are associated with female predominance, mucinous histology, aggressive tumor biology, MSI-H, and CIMP-high tumours.30 BRAF V600E CRCs demonstrate poor response to monotherapy with BRAF inhibitors due to EGFR-mediated feedback activation. However, in a recent BEACON CRC study, dual inhibition of BRAF and EGFR (e.g., encorafenib plus cetuximab) has demonstrated improved clinical outcomes in this subset of metastatic CRC patients.30,31

Non-V600 BRAF mutations (class II and III) are increasingly recognized as relevant molecular subgroups. Class II mutations, such as K601E, demonstrate high kinase activity and signal as RAS-independent dimers. Class III mutations (e.g., D594G) are kinase-impaired and are dependent on upstream RAS activation.32 While these mutations typically confer a better prognosis than V600E, they present unique therapeutic challenges. Experimental inhibitors targeting class II mutations (e.g., pan-RAF inhibitors) show promise in preclinical models, offering potential treatment options for these molecularly distinct CRC subsets.28,30

KRAS-mutated tumors are associated with CMS3 subtype and feature metabolic deregulation.15 Although KRAS mutations, particularly in exons 12 and 13, are considered negative predictive markers for anti-EGFR therapy, therapeutic KRAS G12C inhibitors (e.g., sotorasib, adagrasib) can offer potential benefit in combination with EGFR blockade.28

HER2 (ERBB2) amplification occurs in approximately 2–3 % of metastatic colorectal tumors and occurs mutually exclusively with RAS and BRAF mutations.33 Since these tumors exhibit primary resistance to anti-EGFR monoclonal antibodies, dual HER2 blockade using trastuzumab in combination with agents like lapatinib or pertuzumab has demonstrated favorable clinical outcomes in recent clinical trials.33,34

HER2 IHC/FISH and/or molecular profile testing are essential for identifying patients eligible for HER2-directed therapies. Emerging antibody-drug conjugates targeting HER2, such as trastuzumab deruxtecan, have shown antitumor activity and favorable safety profile as a single agent dose in patients with pretreated HER2-positive metastatic CRC, including those with RAS mutations.35

NTRK1/2/3 fusions are rare in CRC (<1 %), but they are primary oncogenic drivers and occur commonly in MSI-H and RAS/BRAF wildtype tumors. Since these fusions predict sensitivity to tropomyosin receptor kinase (TRK) inhibitors like larotrectinib and entrectinib, their identification is crucial.36 Early-phase studies report response rates exceeding 75 %, and both agents are FDA-approved for treating NTRK fusion-positive cancers regardless of tissue origin.37

i. Tumor Mutational Burden (TMB). TMB measures the nonsynonymous mutations per megabase of tumor DNA and is a surrogate marker for tumours' neoantigenic potential. MSI-H and POLE-mutant tumors are highly responsive to ICIs. FDA-approved regimens include pembrolizumab and nivolumab ± ipilimumab for advanced CRC. Although high TMB is observed in MSI and POLE-mutated tumors and predicts response to immunotherapy, it can also identify a minority subset of MSS CRCs with immunotherapy-sensitive phenotypes.19,25,38,39 The incorporation of immunotherapy in earlier-stage MSI tumors is also currently being explored. The clinical utility of TMB in CRC, outside hypermutated contexts, remains limited, and TMB should not be used as a standalone biomarker in MSS/POLE-wild-type disease. MSS tumors with intermediate TMB consistently respond to PD-1/PD-L1 blockade only in the presence of other immunoregulatory features.40,41 Ongoing trials explore combination strategies involving MEK inhibitors, VEGF inhibitors, or oncolytic viruses with checkpoint inhibitors and agents modulating the tumor microenvironment.42

ii. EGFR Inhibitors. Cetuximab and panitumumab are currently approved for RAS/BRAF wild-type CRCs in combination with chemotherapy, especially effective in left-sided tumours.43,44

iii. BRAF Inhibitors. Single-agent BRAF inhibitors have limited therapeutic significance due to compensatory EGFR signaling. Combination regimens (e.g., encorafenib + cetuximab ± binimetinib) are now standard in BRAF V600E-mutant CRC730.

iv. KRAS G12C. Currently in clinical trials, combination regimens have demonstrated potential in overcoming resistance and expanding KRAS-targeted therapy beyond lung cancer.43,45

v. HER2-Directed Therapy. Trastuzumab combined with pertuzumab or lapatinib is effective and offers response rates >30 % in HER2-amplified, RAS wild-type tumours.35,46,47

vi. NTRK Inhibitors. TRK inhibitors, larotrectinib and entrectinib achieve high response rates and long durations of response and are approved for tumours harbouring NTRK fusions, offering a tissue-agnostic treatment option.36,37,48

Novel emerging strategies include dual inhibition (e.g., MEK + EGFR), synthetic lethality approaches (e.g., PARP inhibitors in HR-deficient tumors), and tumor microenvironment targeting (e.g., TGF-β blockade in CMS4)12,42 in the treatment of CRC.

Recent phase II studies have shown unprecedented complete clinical and pathologic responses to immune checkpoint inhibitors in locally advanced dMMR rectal adenocarcinoma (35,660,797, 38,852,601). These findings suggest neoadjuvant immunotherapy could obviate chemoradiation or surgery in select patients. Pathologists should recognize treatment-related regression patterns to avoid misinterpretation of residual disease (36,791,752).

Circulating tumor DNA (ctDNA) detection has emerged as a sensitive marker of minimal residual disease (MRD) following curative-intent surgery in stage II–III CRC. ctDNA positivity strongly correlates with recurrence risk and may guide adjuvant chemotherapy decisions (40,772,634, 41,115,959). Ongoing studies (e.g., DYNAMIC-III) are evaluating ctDNA-guided escalation and de-escalation strategies to personalize adjuvant management.

Large multiplatform analyses from TCGA established a four‑class framework for gastric and gastroesophageal junction (GEJ) adenocarcinomas that remains the backbone for research and practice: Epstein-Barr virus (EBV)‑associated, MSI, chromosomal instability, and genomically stable.1,49 Crucially, esophageal adenocarcinoma aligns with chromosomally unstable proximal gastric cancers rather than with esophageal squamous cell carcinoma.49 For classification and management, esophageal adenocarcinoma (EAC) belongs on the same molecular map as proximal gastric and GEJ adenocarcinomas50,51 and will collectively be referred to as gastroesophageal carcinoma (GEC) in this review.

Table 3, Table 4 provide a comprehensive list of GEC subtypes based on their molecular or NCCN guideline-driven biomarker profile14 and the corresponding molecular pathology testing and therapeutic implications.

EBV‑positive tumors define a discrete epigenetic and immunologic state. They show pervasive CpG‑island hypermethylation, frequent PIK3CA hotspots, and 9p24.1 gains that include PD‑L1 and PD‑L2.52,53 The microenvironment is lymphocyte‑rich with interferon‑γ signaling and checkpoint overexpression.54 In routine pathology, EBV is identified by EBER in situ hybridization (ISH) on formalin‑fixed tissue, which is sensitive and specific and works on small biopsies.55 EBV positivity is less common at the GEJ compared to the distal stomach,49 yet when present, it immediately places a tumor within an immune‑inflamed, PI3K‑tilted biology that is relevant to PD‑1 therapy and to trials combining PI3K pathway inhibition with immunotherapy.54

MSI‑H cancers are hypermutated and enriched for neoantigens, with dense tumor‑infiltrating lymphocytes and an inflamed transcriptomic profile.56,57 In gastric primaries, MLH1 promoter methylation is a common driver of MMR deficiency.58,59 MSI is strongly predictive of benefit from PD‑1 blockade in advanced disease.60 Accurate calls depend on paired orthogonal methods. MMR IHC provides a rapid first screen that also flags noncanonical patterns. Direct MSI testing by PCR or panel‑based NGS quantifies the fraction of unstable loci.61 Many centers report an NGS score such as MSI sensor with a validated cutoff and add MLH1 methylation when indicated to clarify sporadic vs. hereditary origin.62 MSI‑H tumors occur across GECs and should be recognized early because they shift first‑line therapeutic strategy toward PD‑1-blockade.63

CIN cancers dominate the proximal stomach and GEJ and constitute the defining molecular profile of GECs.49,64 TCGA’s integrated analysis of copy number, exome, transcriptomic, methylation, and proteomic data found that EAC clustered with chromosomally unstable gastric cancers, rather than with esophageal squamous carcinoma.49 The result persisted after removing borderline GEJ cases. The hallmark is aneuploidy with focal high‑level amplifications of receptor tyrosine kinases (RTK) and cell‑cycle drivers (including ERBB2, EGFR, FGFR2, MET, VEGFA, KRAS, CCNE1) on a background of near‑universal TP53 mutation and frequent CDKN2A loss. These focal events are often mutually exclusive and are measurable on copy‑number capable panels.49,64

TCGA also described a proximal‑to‑distal methylation gradient within the CIN class. The most proximal tumors clustered into a hypermethylated group in which MGMT and CHFR are frequently silenced.64 This raises testable hypotheses. MGMT methylation predicts alkylator sensitivity in other cancers. CHFR loss has been linked to taxane sensitivity in preclinical systems and select clinical contexts.65,66 Whether these associations stratify outcome within CIN GECs remains to be proven, yet the gradient highlights real biologic nuance across the esophagus-stomach axis. The epidemiology points in the same direction. Rising reflux and obesity in Western populations parallel the surge of proximal tumors with instability biology, while declining Helicobacter pylori tracks with the relative scarcity of EBV and MSI at the junction compared with the distal stomach.67,68

Genomically stable tumors are best approached as a category of exclusion. After EBV and MSI have been ruled out and copy‑number analysis on an adequate‑purity specimen does not reach CIN thresholds, the remaining cancers fall into this group. Their core biology reflects disruption of cell adhesion and the actin cytoskeletal signaling. Recurring alterations include truncating CDH1 mutations and hotspot mutations in RHOA. CLDN18::ARHGAP fusions appear in a subset and converge on Rho‑GTPase signaling. These tumors are commonly associated with diffuse, poorly cohesive histology and exhibit overall low copy‑number burden. Because they lack many of the canonical targets found in CIN tumors, their molecular profile shifts the focus of therapeutic strategies toward alterative targets, most notably claudin 18.2 (CLDN18.2), thereby redefining what is considered actionable in this subset.51,64

Two caveats can help prevent misclassification. First, the accuracy of copy‑number calling declines with lower tumor purity, so biopsies with <20 % tumor cellularity may appear copy‑number quiet, even when the underlying biology reflects CIN. In such cases, macrodissection or repeat biopsy should be considered, especially when a “genomically stable” designation is discordant with clinical behavior.69 Secondly, quantitative thresholds for defining instability vary by institution and testing platform; these may include panel‑specific clonal deletion scores or defined cut points for the fraction of genome altered. To support consistent interpretation, reports should specify the metric or scoring system used to distinguish stability vs. instability.69, 70, 71

HER2 amplification or overexpression defines a clinically important subset of GECs, especially within the CIN group and those with intestinal‑type histology.72,73 A landmark phase 3 randomized controlled trial established trastuzumab plus chemotherapy as standard first‑line therapy in HER2‑positive advanced gastric and GEJ cancers,74,75 with EAC treated analogously due to shared CIN biology.76

Testing begins with HER2 IHC. Equivocal 2+ HER2 IHC (Fig. 2B) requires ISH confirmation. Intratumoral heterogeneity is common, so multiple biopsies are recommended77 . HER2‑low generally refers to HER2 IHC 1+ or 2+ with negative ISH. Gastric‑specific validation of this subset is ongoing,78 with early data suggesting possible benefit from trastuzumab deruxtecan.79 Until prospective evidence matures, HER2‑low should be regarded as an investigational stratifier rather than a routine treatment label. Reports should state HER2 IHC score, ISH result, and assay clone to enable comparison across trials.77

Trastuzumab plus platinum-fluoropyrimidine chemotherapy remains the first‑line standard in HER2‑positive advanced GECs80 . Upon progression, reassessment of HER2 status is essential. HER2 loss is common after trastuzumab exposure81 and should be evaluated through tissue biopsy when feasible, or via circulating tumor DNA (ctDNA) as an alternative.82 ctDNA can also reveal emergent resistance drivers like KRAS or PIK3CA mutations that blunt antibody-drug conjugate activity.83,84 These practicalities anchor HER2 as both a predictive marker and a dynamic target that requires reassessment over time.

If HER2 remains positive after progression, trastuzumab deruxtecan is an effective second-line agent, though it requires monitor for interstitial lung disease.85 Additional HER2-targeted options under investigation include bispecific antibodies, tyrosine kinase inhibitors, and rational combinations targeting co-altered drivers such as FGFR2 or MET amplifications.86,87 HER-targeted therapy should not be reflexively continued without confirming biomarker persistence before each treatment line.81

In HER2-negative GEC, PD‑L1 CPS stratifies benefit from chemoimmunotherapy80 (Fig. 2C). The CheckMate 649 trial demonstrated the most pronounced survival advantage using nivolumab plus oxaliplatin-fluoropyrimidine chemotherapy in patients with CPS ≥5.88 The benefit extends into lower CPS subgroups, though with attenuation.

Assay harmonization is important. A validated PD-L1 22C3 protocol should be employed, and CPS should be reported alongside the assay clone and platform to reduce variability. Pathology reports are encouraged to include any preanalytic limitations, such as crush artifact or limited tumor content, that may influence the CPS outcome.89,90

Underlying tumor biology and coexisting biomarkers guide immunotherapy use in GEC. First, for HER2-positive tumors, many centers add pembrolizumab to trastuzumab when PD-L1 CPS ≥1, leveraging synergy between HER2-driven oncogenic signalling and immune activation91 . Second, in HER2‑negative metastatic disease, chemoimmunotherapy improves survival, particularly when CPS ≥5. Nivolumab plus oxaliplatin-fluoropyrimidine chemotherapy is a commonly used standard regimen, as shown in the CheckMate 649 trial.92,93 Third, EBV‑positive tumors, despite being microsatellite stable, often show an immune‑inflamed phenotype and may respond to anti-PD‑1 therapy. While prospective randomized data in EBV‑selected cohorts remain limited, the biologic rationale is strong.94,95

MSI‑H status is uncommon but defines a tumor biology with exceptional sensitivity to PD‑1 blockade, warranting immunotherapy (as discussed above) regardless of PD-L1 CPS. Responses are frequent and durable across organ sites, including GECs.96 Routine detection is best achieved through embedding NGS‑based MSI calls, which provides redundancy alongside MMR IHC. When MLH1 is lost, MLH1 promoter methylation testing helps distinguish sporadic from hereditary etiologies. The therapeutic implications are immediate and substantial.62

Claudin 18.2 (CLDN18.2) is a tight‑junction protein normally tucked away within gastric mucosa and aberrantly exposed on malignant cells.97 Expression is common in genomically stable and diffuse-type tumors (Fig. 2D) and less frequent in strongly HER2‑amplified cancers, making it a valuable target in subtypes that historically lacked actionable biomarkers.98

Two phase 3 trials, SPOTLIGHT (with mFOLFOX6)99 and GLOW (with capecitabine plus oxaliplatin),100 showed that adding the monoclonal IgG antibody zolbetuximab to first‑line chemotherapy improved progression‑free and overall survival in CLDN18.2‑positive, HER2‑negative advanced gastric/GEJ cancers disease. Both trials used centralized IHC with the Ventana CLDN18 (43–14A) RxDx assay101 and defined CLDN18.2 positivity as moderate to strong membranous staining in ≥75 % of tumor cells, a threshold met by approximately 38 % of tumors in these cohorts. Reliable implementation requires validated CLDN18.2 IHC assays within each individual laboratory. A Global Ring study has confirmed the reproducibility of CLDN18.2 IHC across multiple commercially available platforms.102

Zolbetuximab combined with oxaliplatin‑based chemotherapy is now an evidence‑based first-line option for CLDN18.2‑positive, HER2‑negative tumors.99,100 Where both HER2 and CLDN18.2 are co-expressed, HER2 is typically prioritized due to more mature survival data and subsequent treatment options. When HER2 is negative, CLDN18.2 becomes the primary target.103

Ongoing studies are evaluating multiple types of anti-CLDN18.2 agents, including bispecific antibodies.104,105 Combinations such as zolbetuximab plus PD‑1 blockade are also under investigation, which may further improve outcomes.104,106

LINE‑1 encodes ORF1p, an RNA‑binding protein largely absent from normal tissues and broadly expressed in epithelial cancers.107,108 ORF1p can be detected in both tissue and in plasma using ultrasensitive assays, and early studies suggest that circulating protein correlates with tumor burden and treatment response.109,110 While not yet in clinical use, it represents a promising non-invasive biomarker that complements ctDNA in settings where serial biopsy is limited.111