Tumor-node response heterogeneity: Exploring the “spatial biology” of neoadjuvant therapy response in rectal cancer and its clinical implications

Abstract

Neoadjuvant chemoradiotherapy - the combination of chemotherapy and radiation therapy administered before surgery - followed by total mesorectal excision is the foundation of treatment for locally advanced rectal cancer. Pathological assessment of the primary tumor commonly uses tumor regression grading, which quantifies the extent of tumor response to treatment and provides prognostic value. However, a growing body of evidence reveals profound inter- and intra-patient heterogeneity in treatment response between the primary tumor and its associated lymph nodes. This review explores the concept of ‘spatial biology’ - the study of how distinct anatomical compartments within a patient, such as the primary tumor site and nodal basins, show unique biological behaviors and distinct responses to therapy. We will synthesize current evidence from single-cell and spatial transcriptomic studies, explore the dynamic remodeling of the tumor immune microenvironment at the primary site, and contrast this with the unique (and often discordant) response within lymph nodes that may serve as immunological sanctuaries for resistant clones. The combination of a novel lymph node regression grading system, which specifically assesses treatment response in lymph nodes, with traditional tumor regression grading is proposed here to more accurately reflect this spatial variation. We will critically appraise current evidence supporting the use of lymph node regression grading, discuss its limitations and difficulties for clinical implementation, and propose a framework for risk stratification informed by spatial biology to guide tailored adjuvant strategies. Finally, we will describe potential future research directions, including multi-omics spatial mapping, circulating tumor DNA integration, and novel clinical trial designs incorporating lymph node-sparing radiotherapy. This system acts as a foundation for individualized precision oncology in rectal cancer.

Key Words: Rectal cancer; Lymph node regression grade; Spatial biology; Tumor microenvironment; Precision oncology

Core Tip: This review article presents a spatial biology framework that explains why primary tumors and metastatic lymph nodes in locally advanced rectal cancer often respond differently to neoadjuvant chemoradiotherapy. Key points include clarifying the mechanisms of immune evasion and mitochondrial genome dynamics, and integrating tumor and lymph node regression grades to define spatial response phenotypes. The framework enhances prognostic stratification, as evident in the nodal sanctuary phenotype, and facilitates tailored therapeutic strategies through a translational roadmap that includes circulating tumor genetic material-guided therapy and lymph node-sparing radiotherapy trials. Ultimately, this approach aims to advance compartment-aware, biology-driven precision oncology in rectal cancer.

INTRODUCTION

Neoadjuvant chemoradiotherapy (nCRT) has changed how we manage locally advanced rectal cancer (LARC). It helps downstage tumors, makes them easier to remove, and destroys micrometastatic disease. The standard measure of treatment effect has focused on the primary tumor. Tumor regression grading (TRG) systems are used for this purpose[1]. However, new evidence reveals a worrying pattern. Many patients have different pathological responses in the primary tumor and lymph nodes[2,3]. For example, a patient may have a nearly complete response in the rectum, but still have viable tumor deposits in the mesorectal lymph nodes, or the reverse.

This observation signals the core limitation of viewing a tumor as a single entity. Instead, it argues for a ‘spatial biology’ framework, where the primary tumor and its nodal metastases stand as distinct though interconnected ecosystems that may respond independently to systemic pressures such as nCRT[4]. In this context, we operationally define ‘spatial biology’ as the study of how the anatomically distinct microenvironments of the primary tumor and its metastatic lymph nodes (each having unique immunological, stromal, and vascular features) drive divergent therapeutic responses and shape clinical outcomes. A recent study introduced and validated a novel lymph node regression grading (LRG) system. It supplies key clinical evidence for this regional heterogeneity and demonstrates that LRG holds independent prognostic power, particularly in node-positive disease[5].

This review attempts to deliver a comprehensive overview of the current understanding of tumor-node response heterogeneity in rectal cancer. Specifically, we will: (1) Synthesize the biological basis of this discordance by drawing on recent advances in single-cell and spatial transcriptomics; (2) Systematically assess the emerging role of LRG as a prognostic biomarker; (3) Discuss the clinical implications of risk stratification and adjuvant therapy; (4) Outline a roadmap for translating spatial biology into clinical practice; and (5) Identify key future research directions to promote the field toward personalized precision oncology.

PRIMARY TUMORS VS LYMPH NODE MICROENVIRONMENTS

Tumor response to nCRT is not just cell-autonomous; it is controlled by the stromal and tumor immune microenvironment. The microenvironments of primary colorectal mucosa and metastatic lymph nodes differ in structure and function, resulting in different responses. This spatial separation underlies treatment response heterogeneity, recently shown by ultra-high-resolution profiling technologies[6-8].

Primary tumor microenvironment

At the primary tumor site, nCRT induces a complex cascade of events. Radiation and chemotherapy induce immunogenic cell death, which triggers the release of tumor-associated antigens and damage-associated molecular patterns[9]. This can, in principle, stimulate a potent anti-tumor immune response characterized by the infiltration of CD8+ cytotoxic T cells and a change towards a pro-inflammatory cytokine milieu[10]. Emerging single-cell and spatial transcriptomic research is producing unmatched resolution into this active remodeling. Xu et al[11] constructed comprehensive atlases of colorectal cancer, revealing the distinct transcriptional programs of epithelial cells across disease stages and identifying SCAN-domain-containing protein 1 as a novel regulator of malignant epithelial plasticity. Notably, SCAN-domain-containing protein 1 overexpression was shown to attenuate T cell-mediated cytotoxicity, thus representing an additional immune-evasive role. Bakke et al[12] also demonstrated that the dissemination of mitochondrial DNA (mtDNA) variants via extracellular vesicles reflects the metabolic state of the tumor microenvironment (TME), in which high cellular density may impede the clearance of pathogenic mtDNA variants, consequently promoting tumor aggressiveness. Zhang et al[13] provided a comprehensive single-cell atlas of colorectal cancer, discovering distinct fibroblast subtypes that differentially modulate immune infiltration and therapeutic response. Figure 1 depicts the dynamic (and often paradoxical) remodeling of the post-nCRT tumor immune microenvironment.

Figure 1 Dynamic remodeling of the primary tumor immune microenvironment following neoadjuvant chemoradiotherapy. Neoadjuvant chemoradiotherapy or total neoadjuvant therapy initiates an antitumor immune response by inducing immunogenic cell death, which releases tumor-associated antigens and damage-associated molecular patterns. This process has been shown to stimulate CD8+ T-cell expansion, M1 macrophage polarization, and enhanced antigen presentation. Recent single-cell studies have identified a positive feedback loop between Interferon-gamma+ effector memory T cell and atypical chemokine receptor 1+ endothelial cells that potentiates antitumor immunity. However, neoadjuvant chemoradiotherapy simultaneously activates robust immunosuppressive pathways, including transforming growth factor-β-driven fibrosis, cancer-associated fibroblast activation, regulatory T cell and myeloid-derived suppressor cell recruitment, and SCAN-domain-containing protein 1-mediated immune evasion. Mitochondrial DNA dynamics add further to metabolic reprogramming and tumor aggressiveness. The equilibrium between these opposing forces is a pivotal factor in determining the tumor regression grade, thereby signifying the spatial variation of the primary tumor response. nCRT: Neoadjuvant chemoradiotherapy; TNT: Total neoadjuvant therapy; TAAs: Tumor-associated antigens; DAMPs: Damage-associated molecular patterns; TIME: Tumor immune microenvironment; IFNG: Interferon-gamma; EC: Endothelial cell; TRG: Tumor regression grade; SCAND1: SCAN-domain-containing protein 1; MDSC: Myeloid-derived suppressor cell; TGF: Transforming growth factor; CAF: Cancer-associated fibroblast; mtDNA: Mitochondrial DNA; Tem: Effector memory T cell; Treg: Regulatory T cell.

However, nCRT simultaneously activates robust profibrotic and immunosuppressive pathways. In rectal cancer, a hallmark is the induction of dense stromal fibrosis, driven by the activation of cancer-associated fibroblasts and transforming growth factor-beta signaling[14]. This fibrotic stroma creates a physical barrier that sequesters effector immune cells and promotes hypoxia, continuing to drive treatment resistance[15]. Moreover, nCRT upregulates immune checkpoint molecules such as programmed death ligand-1 on both tumor and immune cells, and recruits regulatory T cells and myeloid-derived suppressor cells, thereby suppressing initial immune activation[16,17]. Notably, Le Saux et al[18] used multiplex immunofluorescence to demonstrate that the spatial distribution of programmed death ligand-1+ cells relative to CD8+ T cells is a key determinant of nCRT response, as greater spatial distance is associated with treatment resistance. Thus, the balance between these opposing forces - immunogenic cell death vs stromal/immune suppression - ultimately determines the TRG at the primary site. Liu et al[19] further showed that the composition and spatial organization of immune cells within the TME are key determinants of post-nCRT recurrence risk.

Lymph nodes as compartmental sanctuaries

Lymph nodes are not simply passive places for metastatic cells. They are highly organized immune organs. Their response to nCRT is even more complex because of their normal function. Tumor spread to a lymph node changes its structure and immune environment. Tumor cells may cause fibrosis, disrupt lymphatic flow, and alter the presence and function of immune cells[20].

Recent single-cell and spatial transcriptomic studies have begun to reveal that metastatic lymph nodes frequently harbor a more profoundly immunosuppressive microenvironment than the primary tumor, marked by higher ratios of regulatory T cells and exhausted T cell phenotypes[21]. A landmark study by Zhou et al[22] performed single-cell RNA sequencing of paired primary tumors and metastatic lymph nodes in colorectal cancer to reveal that lymph node metastases exhibit a distinct immune landscape with an enrichment of lymphocyte activation gene 3+ exhausted T cells and secreted phosphoprotein 1+ macrophages, indicating the unique selective pressures that the nodal microenvironment imposes on metastatic clones. This suggests that lymph nodes may provide a protective niche for metastatic clones. Therapeutic doses of radiation to nodal basins, while cytotoxic, may also intensify this fibrotic and immunosuppressive remodeling, potentially creating a ‘sanctuary’ effect whereby residual tumor cells are shielded[23]. Figure 2 outlines the spectrum of lymph node responses to nCRT from sterilization to progression.

Figure 2 Spectrum of lymph node responses to neoadjuvant chemoradiotherapy and the biological significance of lymph node regression grade. Metastatic lymph nodes exhibit three primary patterns of response to neoadjuvant chemoradiotherapy: (1) Complete clearance lymph node regression grade (LRG1), characterized by fibrous scarring or acellular mucin lakes; (2) Partial regression (LRG2-5), with residual tumor cells embedded in a fibrotic stroma; and (3) Intrinsic resistance (LRG4-5), where tumor cells show almost no regression. The outcomes observed are the result of a combination of mechanisms, including immune clearance, treatment-induced senescence, fibrous encapsulation, and the selection of resistant clones by the immune microenvironment. lymph node regression grade, when utilised as a histological indicator, offers a direct reflection of the resistance capacity of the lymph node microenvironment. Its grade has been shown to be closely associated with patient prognosis. nCRT: Neoadjuvant chemoradiotherapy; LRG: Lymph node regression grade.

This intrinsic biological difference explains why primary tumors (surrounded by a distinct cytokine and stromal milieu) may respond favorably, while nodal deposits, entrenched in a fibrotic, immunosuppressive niche, persist. The LRG score is therefore a direct histological readout of the nodal microenvironment’s potential chemoresistance[5,24].

Other potential mechanisms of response discordance

Although these immune and stromal factors are central, the differential response phenomenon is likely multifactorial. Pre-existing clonal heterogeneity between the primary tumor and its associated lymph node metastases may lead to divergent sensitivity to chemotherapy and radiation. Whole exome sequencing studies have demonstrated that lymph node metastases can exhibit distinct mutational profiles compared to the primary tumor, including mutations in genes associated with DNA damage repair that may confer differential sensitivity to nCRT[25,26]. Differences in vascularization and drug penetration between the highly vascularized primary tumor bed and the less permeable nodal tissue may likewise contribute to this discordance. Variability in radiation dose distribution within the pelvis and the development of treatment protocols towards total neoadjuvant therapy (TNT) - which may differentially impact these compartments - introduce further complexity. Han et al[27] performed a comprehensive review of how various TME components (including T cells, macrophages, cancer-associated fibroblasts, and the extracellular matrix) modulate therapeutic resistance in colorectal cancer. Li et al[28] used integrated single-cell and spatial transcriptomics to identify a glycolysis-enriched tumor subpopulation in microsatellite instability-high colorectal cancer that colocalizes with immunosuppressive niches. This demonstrates how tumor cell-intrinsic metabolic states can shape the local immune landscape and contribute to therapeutic resistance. Kloesch et al[29] further demonstrated that metabolic heterogeneity between primary tumors and metastases contributes to differential responses to immunotherapy.

LRG: ITS EVIDENCE, LIMITATIONS, AND CLINICAL POTENTIAL

The LRG scoring system constitutes a significant advance in post-nCRT pathological assessment. Originally proposed by Caricato et al[30] and then refined by Shi et al[5] and Ozturk et al[24], LRG provides a semiquantitative assessment of treatment response within metastatic lymph nodes based on the proportion of residual viable tumor cells relative to fibrotic cells.

The prognostic value of LRG

Multiple studies have validated LRG’s prognostic significance. In a large retrospective cohort of 469 LARC patients, Ozturk et al[24] demonstrated that LRG was a significant predictor of disease-free survival independent of pathological lymph node stage, with patients attaining complete nodal response (LRG1) exhibiting outcomes comparable to those of pathological lymph node stage 0 patients. Shi et al[5] further showed that LRG is an independent predictor of both mortality and recurrence in patients with node-positive disease, with hazard ratios of 1.533 and 1.278, respectively, after multivariate adjustment. Importantly, the prognostic value of LRG appears to be additive to that of TRG, suggesting that these two scores capture non-redundant biological information. Athauda et al[31] conducted a pooled analysis of two esophagogastric cancer trials and found that combining the primary and nodal regression scores provided superior risk stratification compared to either score alone. Zhuang et al[32] recently extended this concept to lateral lymph nodes, demonstrating that LRG in lateral nodes carries independent prognostic significance and may help guide clinical decision-making regarding lateral lymph node dissection.

Limitations and difficulties

Despite its promise, the medical adoption of LRG confronts multiple challenges. The most significant is the difficulty in telling apart between LRG0 (uninvolved node) and LRG1 (sterilized metastasis), since both can appear as fibrotic nodules on routine histology[24]. This calls for exact pathological examination, potentially including extensive serial sectioning of lymph nodes from areas with high pre-test probability of metastasis. The interobserver reproducibility of the 6-tier LRG system also requires confirmation across multiple institutions before it can be standardized. In addition, the optimal threshold for defining ‘high-risk’ nodal response remains to be established. While some studies have used LRG4-5 as the cut-off, others have proposed alternative thresholds. Prospective studies are therefore needed to determine the most clinically meaningful threshold[33]. Figure 3 illustrates this spatial biology-to-precision medicine pipeline.

Figure 3 Spatial biology-informed precision oncology treatment pipeline for locally advanced rectal cancer. The integration of primary tumor regression grade (TRG) and lymph node regression grade (LRG) enables the classification of patients with locally advanced rectal cancer into four distinct spatial response phenotypes: (1) Coordinate complete response (TRG1 + LRG1 or pathological lymph node stage 0); (2) Nodal sanctuary phenotype (TRG1-2 + LRG4-5); (3) Primary resistance (TRG4-5 + LRG1-2); and (4) Coordinate resistance (TRG4-5 + LRG4-5). Each distinct phenotype is associated with a unique therapeutic strategy. Patients exhibiting a coordinated complete response may be considered for treatment de-escalation, encompassing active surveillance or the omission of adjuvant chemotherapy. The nodal sanctuary phenotype is characterised by discordant response, whereby the primary tumor responds well but nodal disease persists. This requires escalation of treatment, including intensified adjuvant chemotherapy, combination immunotherapy, a radiotherapy boost to residual nodes, or emerging lymph node-sparing radiotherapy approaches. Primary resistance and coordinate resistance phenotypes mandate enrolment in novel clinical trials exploring targeted therapy or immunotherapy combinations, frequently guided by circulating tumor DNA dynamic monitoring. Emerging approaches, including circulating tumor DNA-guided therapy, lymph node-sparing radiotherapy and spatial multi-omics profiling, are currently being validated in clinical trials, suggesting potential for further development of the current precision oncology framework. LARC: Locally advanced rectal cancer; nCRT: Neoadjuvant chemoradiotherapy; ctDNA: Circulating tumor DNA; TRG: Tumor regression grade; LRG: Lymph node regression grade; RT: Radiotherapy.

Table 1 summarizes the representative prospective studies on the spatial biology of malignant rectal tumors, including research on circulating tumor DNA (ctDNA)-guided therapy, lymph node-sparing radiotherapy, and immunotherapy combinations. Taken together, these studies represent a translational pipeline for merging spatial biology into clinical practice.

Table 1 Representative prospective studies in spatial biology and precision oncology for colorectal cancer.

Ref.	Phase	Population	Strategy	Endpoint	Status
Yao[38]	Prospective	LARC post-nCRT	ctDNA-guided adjuvant therapy decision	2-year DFS	Recruiting
University of Florida[39]	Phase II	ERC	Post-TNT ctDNA-guided watchful waiting vs surgery	Organ preservation rate	Recruiting
Mögele et al[37]	Prospective	LARC	ctDNA for prediction of pCR after nCRT	pCR prediction accuracy	Completed
Song et al[45]	Phase II	MSS ERC	Node-sparing short-course RT + CAPOX + PD-1/CTLA-4	pCR rate	Completed
Song et al[45]	Phase III	MSS LARC	Node-sparing short-course RT + CAPOX + tislelizumab vs CRT	3-year DFS	Completed
Bando et al[42]	Phase II	LARC	nCRT + nivolumab	pCR rate	Completed
Mo et al[51]	Prospective	Stage II-III CRC	ctDNA-MRD detection for recurrence risk stratification	2-year RFS	Completed
Tie et al[35] and Tie et al[36]	Phase II/III	LARC	ctDNA-guided adjuvant therapy de-escalation	Non-inferiority in RFS	Ongoing

LARC: Locally advanced rectal cancer; nCRT: Neoadjuvant chemoradiotherapy; ctDNA: Circulating tumor DNA; DFS: Disease-free survival; ERC: Early rectal cancer; TNT: Total neoadjuvant therapy; pCR: Pathological complete response; MSS: Microsatellite stable; RT: Radiotherapy; CAPOX: Capecitabine and oxaliplatin; PD-1: Programmed cell death protein 1; CTLA-4: Cytotoxic T-lymphocyte-associated protein 4; CRT: Chemoradiotherapy; CRC: Colorectal cancer; MRD: Minimal residual disease; RFS: Recurrence-free survival.

THE CLINICAL IMPERATIVE: INTEGRATING SPATIAL BIOLOGY INTO RISK STRATIFICATION

Heterogeneity has direct clinical implications. The classic linear model of using the primary tumor TRG (or post-therapy pathologic T stage) alone for prognostic stratification is insufficient. A new, integrated spatial risk assessment model is needed.

From tumour, node, and metastasis to spatial risk phenotyping

The current post-neoadjuvant post-therapy pathologic tumour, node, and metastasis staging system precisely captures disease location and burden, though it neglects its biological characteristics. The combined use of TRG and LRG provides a powerful ‘spatial response phenotype’[5,31] which captures four scenarios: (1) Primary resistance phenotype (pow TRG, any LRG): The primary tumor is intrinsically resistant, thus driving poor prognosis; (2) Nodal sanctuary phenotype (high TRG, low LRG): The paradigmatic discordant response. A strong primary response can mask persistent nodal disease - a scenario in which adjuvant therapy is likely crucial but may be easily overlooked based on primary tumor assessment alone[3,34]; (3) Coordinate resistance phenotype (low TRG, low LRG): Systemic resistance across both compartments indicates a requirement for a comprehensive treatment strategy reevaluation; and (4) Coordinate complete response phenotype (high TRG, high LRG/pathological lymph node stage 0): Excellent prognosis; potential candidates for treatment de-escalation. The synergistic prognostic value of TRG and LRG suggests that they capture non-redundant biological information. LRG may be a superior marker of the host immune system’s ability to clear disseminated disease - an important determinant of far failure[35].

Guiding individualized adjuvant therapy

This spatial biology framework directly informs the ‘adjuvant question’. For patients exhibiting the high-risk ‘nodal sanctuary’ phenotype, treatment escalation is warranted. This may include: (1) Intensified chemotherapy: Utilizing doublet or triplet regimens in an adjuvant setting and using ctDNA positivity to monitor efficacy. A randomized dynamic trial has established the feasibility of using ctDNA-guided adjuvant therapy in stage II colon cancer, and an ongoing dynamic-III trial is extending this approach to locally advanced colon cancer[35,36]. The NEORECT trial has proved the feasibility of ctDNA-based personalized monitoring in rectal cancer by detecting unique ctDNA patterns associated with treatment response[37]. Clinical trials such as ctDNA-minimal residual disease-rectal cancer are currently conducting prospective studies to validate multi-omics models that integrate nucleic acid mutations, copy number variations, and mtDNA to predict the treatment efficacy of neoadjuvant therapy[38]. The ULTIMATE trial is exploring ctDNA-informed management of early-stage rectal cancer, using post-TNT ctDNA results to guide clinical decision-making towards either watchful waiting or standard surgery[39]; (2) Integration of immunotherapy: For mismatch repair-deficient tumors, adjuvant immunotherapy is standard. For proficient mismatch repair (pMMR) tumors that are immunologically ‘cold’, strategies to address nodal immunosuppression are being investigated. These include the combination of nCRT with immune checkpoint inhibitors, transforming growth factor-beta inhibitors (to reduce fibrosis), or compounds targeting myeloid cells[40,41]. Recent phase II trials investigating the efficacy of neoadjuvant immunotherapy combined with nCRT in pMMR rectal cancer have had promising results, with pathological complete response rates ranging between 20%-30%. However, optimal patient selection remains an active area of investigation[42,43]; and (3) Novel targeted radiotherapy: Techniques such as delivering magnetic resonance imaging (MRI) linear-accelerator-guided boosts to poorly responding lymph nodes are currently being investigated to overcome physical sanctuary effects[44]. Even more radical radiotherapy strategies are emerging, such as ‘lymph node-sparing’ in which radiation zones are intentionally limited to the primary tumor site to preserve the immune structures within lymph nodes, followed by immunotherapy. Building on promising phase II results, the mRCAT-III trial is comparing node-sparing modified short-course radiotherapy combined with capecitabine and oxaliplatin and tislelizumab vs conventional short-course chemoradiotherapy in locally advanced pMMR rectal cancer[45]. These trials could fundamentally change how we conceptualize treatment across nodal compartments.

Conversely, the rationale for adjuvant chemotherapy is weak for patients exhibiting the ‘coordinate complete response’ phenotype (TRG1 and LRG1). These patients are ideal candidates for active surveillance or treatment de-escalation trials, thereby avoiding unnecessary toxicity[46].

A ROADMAP FOR SPATIAL PHENOTYPING TO CLINICAL PRACTICE

Translating the spatial biology framework into clinical practice requires a multi-pronged approach addressing standardization, validation, and integration alongside emerging technologies.

Standardizing pathological assessment

The first step in this roadmap is to establish uniform protocols for conducting LRG assessment. The 6-tier system proposed by Shi et al[5] constitutes a useful framework. However, interobserver variability must be addressed by structured training programs and the formulation of consensus guidelines. Digital pathology and artificial intelligence-assisted image analysis may improve repeatability and reduce interobserver variability[47,48]. The distinction between LRG0 and LRG1 also calls for careful logical examination. To this end, extensive serial sectioning of lymph nodes from areas with high pre-test probability of metastasis may improve the detection of occult residual disease[24].

The combination of molecular biomarkers

LRG should not be used in isolation; rather, it should be integrated with other biomarkers to develop composite risk models. ctDNA is especially promising in this regard. The presence of ctDNA following nCRT is strongly associated with high recurrence risk. Therefore, combining LRG with ctDNA status could allow more precise risk stratification[37,49]. Emerging approaches such as methylated ctDNA analysis and fragmentation pattern analysis may further enhance sensitivity for detecting minimal residual disease[50,51].

Table 2 presents the biological characteristics, molecular features, and therapeutic implications of the four spatial response phenotypes derived from combined TRG and LRG assessment. Each phenotype is associated with distinct immune and stromal features and possible therapeutic targets, which correlate with specific strategies currently being explored in clinical trials. This system delivers a roadmap for transitioning from histopathological assessment to mechanism-guided personalized therapy.

Table 2 Mechanisms and translational strategies for spatial biology-informed precision oncology in rectal cancer.

Spatial phenotype	Biological characteristics	Immune features	Potential therapeutic targets	Clinical trial
Coordinate complete response (TRG1 + LRG1/ypN0)	Robust antitumor immunity; immune-mediated clearance; minimal residual disease	Enriched IFNG+ CD8+T; ACKR1+ EC cross-talk; low Treg/MDSC infiltration; low CAF activity	Immune checkpoint preservation; avoid overtreatment	Watch/wait protocols; de-escalation trials[35,36,39]
Nodal sanctuary phenotype (TRG1-2 + LRG4-5)	Primary tumor responds; nodal metastases persist; immunosuppressive nodal niche	Nodal Treg enrichment; exhausted T-cell phenotype; secreted phosphoprotein 1+ macrophages; fibrotic encapsulation	Immune checkpoint inhibitors; TGF-β inhibitors; nodal radiotherapy boost; lymph node-sparing RT	mRCAT-E; mRCAT-III; nodal boost trials[42,45]
Primary resistance (TRG4-5 + LRG1-2)	Primary tumor resistant; nodal clearance achieved; primary TME immunosuppression	SCAND1-mediated immune evasion; mtDNA metabolic reprogramming; CAF-enriched stroma; hypoxia	SCAND1 targeting; metabolic modulators; CAF reprogramming; anti-angiogenic agents	Combination IO; novel targeted therapy trials[11,12]
Coordinate resistance (TRG4-5 + LRG4-5)	Systemic resistance; both compartments refractory; poor prognosis	Clonal heterogeneity; multi-compartment immunosuppression; high mutational burden	Intensified systemic therapy; novel IO combinations; ctDNA-guided adaptive therapy	DYNAMIC-rectal; FOLFOXIRI-based intensification[35,36,39]

TRG: Tumor regression grade; LRG: Lymph node regression grade; ypN: Pathological lymph node stage; IFNG: Interferon-gamma; ACKR1: Atypical chemokine receptor 1; EC: Endothelial cell; Treg: Regulatory T cell; MDSC: Myeloid-derived suppressor cell; CAF: Cancer-associated fibroblast; TGF-β: Transforming growth factor-beta; RT: Radiotherapy; mRCAT: Node-sparing radiotherapy trial; TME: Tumor microenvironment; SCAND1: SCAN-domain-containing protein 1; mtDNA: Mitochondrial DNA; IO: Immunotherapy; ctDNA: Circulating tumor DNA; FOLFOXIRI: Fluorouracil, leucovorin, oxaliplatin, and irinotecan.

Novel trial designs

Future clinical trials should stratify patients based on combinatory TRG-LRG phenotypes. For patients with the ‘nodal sanctuary’ phenotype, randomized trials comparing standard adjuvant chemotherapy vs intensified regimens or novel immunotherapy combinations are warranted. For patients with the ‘coordinate complete response’ phenotype, de-escalation trials could be conducted to evaluate whether adjuvant chemotherapy can be safely omitted. Ongoing ‘lymph node-sparing’ radiotherapy trials represent a major shift toward preventing nodal failure and are likely to alter the role of radiotherapy in early-stage and locally advanced disease[45].

FUTURE DIRECTIONS: MAPPING THE SPATIAL LANDSCAPE

Future paradigms will move beyond histology toward the multi-omics mapping of different spatial compartments. Several emerging technologies and approaches are driving this shift.

Multi-omics spatial profiling

Pre- and post-nCRT biopsies of both primary tumors and lymph nodes (via imaging-guided sampling), analyzed using single-cell RNA sequencing, multiplex immunofluorescence, and spatial transcriptomics, reveal the exact molecular circuits driving response and resistance in each site[52,53]. Avraham-Davidi et al[54] combined scRNA-seq and Slide-seq to create a detailed spatial map of colorectal cancer. The map revealed that tumors are organized inside cellular neighborhoods with distinct compositions and local interactions. The study also demonstrated that these organizational features are conserved between mouse models and human disease, and correlate with clinical outcomes. This approach can be used to identify novel, compartment-specific treatment targets.

Radiomics and non-invasive monitoring

Non-invasive spatial monitoring via ‘radiomics’ (the extraction of multidimensional data from pre- and post-treatment MRI scans) shows promise for predicting both TRG and LRG preoperatively, enabling earlier therapeutic intervention[55,56]. Deep learning approaches applied to baseline and restaging MRI scans have demonstrated high accuracy for predicting pathological responses, and may eventually enable non-invasive assessment of nodal response without the need for surgical sampling[57,58].

Functional imaging and metabolic profiling

Advanced positron emission tomography (PET) imaging with novel tracers may provide insight into the metabolic states of responsive vs resistant compartments. Although fluorodeoxyglucose-PET remains the standard, new immune-cell-targeting tracers or specific metabolic pathways may enable the non-invasive characterization of the spatial biology of treatment response[59,60]. The combination of PET with MRI may provide complementary information about both metabolic function and anatomical response[61].

CONTROVERSIES AND UNSETTLED QUESTIONS

Despite the growing body of evidence supporting the spatial biology framework, several controversies and pending questions remain that warrant discussion.

The clinical readiness of LRG

While LRG has been validated in multiple retrospective cohorts, prospective validation is still lacking. The interobserver reproducibility of the 6-tier system has not been formally assessed in multi-center studies. Furthermore, the optimal cut-off for defining ‘high-risk’ nodal response is still debated. Some studies have used LRG4-5 as the threshold, while others suggest that any residual viable tumor cells (LRG2-5) confer increased risk[5,24]. Up to these questions are addressed through prospective studies, LRG should be considered an investigational biomarker rather than a practice-changing tool.

Treating nodes vs treating the patient

A fundamental question raised by the spatial biology framework is whether we should target specific compartments (e.g., with nodal radiotherapy boost) or adopt systemic strategies that address both compartments simultaneously. The emerging ‘lymph node-sparing’ radiotherapy approach elects the former strategy to preserve nodal immune structures while intensifying systemic therapy. The latter approach (intensified systemic therapy with standard radiotherapy) may be equally valid. Head-to-head comparisons are needed to determine which strategy optimizes clinical outcomes[45].

The combination of TNT

The shift toward TNT has changed the landscape of LARC treatment, leading to higher complete response rates. However, this also raises new questions regarding response assessment. Preliminary data suggest that the spatial biology framework remains applicable in the TNT era, though response dynamics may differ. Studies evaluating the use of TRG and LRG in patients receiving TNT are needed to validate the spatial biology framework in this new treatment paradigm[62,63].

CONCLUSION

The era of evaluating rectal cancer response to nCRT through the lens of the primary tumor exclusively is quickly evolving. The strong evidence for tumor-node response heterogeneity calls for the adoption of a ‘spatial biology’ paradigm[64-66]. The primary tumor and its lymph nodes are biologically distinct arenas in which therapy either succeeds or fails through individual mechanisms. The combination of LRG and TRG is a vital first step toward clinically operationalizing this concept to afford a more detailed and accurate prognostic assessment.

To translate this understanding into improved clinical outcomes, we must: (1) Standardize and validate the pathological assessment of LRG across multiple centers; (2) Design clinical trials that adopt the spatial response phenotypes (through the combination of TRG and LRG) as stratification factors for adjuvant therapy escalation or de-escalation; and (3) Conduct translational research employing multi-omics spatial profiling to decipher the root causes of discordant responses and identify novel therapeutic vulnerabilities in ‘nodal sanctuary’ microenvironments. The strategies proposed here are currently investigational and require prospective validation before they can become practice-changing. Nevertheless, they offer an intellectually rigorous and clinically promising roadmap for further development in this field. Adopting the spatial differences patterns of rectal cancer will enable a defining move towards an era in which treatment is not based on population averages but is precisely designed to the unique biological landscape of each patient’s disease.