Tailoring therapy through response evaluation: A new era in the management of locally advanced rectal cancer

Abstract

Neoadjuvant chemoradiotherapy combined with surgical resection is considered standard care for locally advanced rectal cancer (LARC). Post-neoadjuvant chemoradiotherapy evaluation and tumor restaging are pivotal components of the treatment paradigm for LARC. These two components not only determine the type of subsequent treatment strategies but also provide the basis for prognosis prediction. Thus, they serve as the cornerstone for accurate and effective patient management in LARC. Currently, conventional assessment modalities, such as imaging and endoscopy evaluation, are widely utilized in clinical practice. At the same time, continuous breakthroughs in emerging technologies, such as circulating tumor DNA detection, multi-omics techniques, artificial intelligence-assisted decision-making, intestinal flora, and patient-derived organoids, provide new perspectives on restaging in LARC, which is expected to make up for the limitations of traditional methods. Additionally, novel therapeutic modalities, such as total neoadjuvant therapy, immune checkpoint inhibitors, and watch-and-wait strategies, impose higher demands for accurate treatment response evaluation and restaging. An in-depth review and discussion of the relevant assessment methods, optimal timing, standardized strategies, and cutting-edge advances hold significant clinical relevance and scientific value in the treatment of LARC. Specifically, it may improve the clinical diagnosis and treatment of LARC, optimize individualized therapeutic decision-making, and improve the prognosis of patients.

Key Words: Rectal cancer; Neoadjuvant therapy; Response evaluation; Tumor restaging; Minireview

Core Tip: Neoadjuvant chemoradiotherapy followed by surgery is the standard treatment for locally advanced rectal cancer. Accurate post-treatment evaluation and restaging are crucial to guide subsequent treatment and prognosis prediction. While conventional methods such as imaging and endoscopy are widely used, their application is hampered by several limitations. Emerging technologies such as detection of circulating tumor DNA, multi-omics, and artificial intelligence offer new perspectives for cancer restaging. Novel therapies such as total neoadjuvant therapy and watch-and wait strategies require enhanced precision in response assessment to optimize individualized therapeutic decisions and improve prognosis in patients with locally advanced rectal cancer.

INTRODUCTION

Rectal cancer is one of the most malignant gastrointestinal tumors, with a high incidence worldwide. The disease burden is particularly heavy in China. According to the latest epidemiological data released by the National Cancer Center of China, colorectal cancer had the second-highest incidence rate and the fourth-highest mortality rate among malignant tumors in the country in 2022[1], with rectal cancer accounting for about half of all reported cases[2], posing a serious threat to national health. Locally advanced rectal cancer (LARC) is defined as cT3-T4 stage tumors or lymph node-positive tumors without distant metastases[3]. Traditional standard treatment modalities for LARC include neoadjuvant chemoradiotherapy (nCRT) followed by total mesorectal excision. Subsequent adjuvant chemotherapy is then administered based on post-surgery pathological staging[4]. In recent years, new strategies such as total neoadjuvant therapy (TNT) and immunotherapy have been gradually integrated into the perioperative treatment of LARC[5]. Meanwhile, emerging technologies, such as the use of circulating tumor DNA (ctDNA), multi-omics analysis, artificial intelligence (AI)-assisted decision-making, intestinal flora modulation, and the use of patient-derived organoid (PDO) models, continue to make this field more precise. Efficacy evaluation and restaging after neoadjuvant therapy (NAT) not only determine the optimization of subsequent treatment strategies but also closely correlate with long-term overall survival, disease-free survival, and postoperative quality of life of patients. Therefore, a systematic exploration of the evaluation methods, timing, standardization system, and application of emerging technologies for the evaluation and restaging for post-neoadjuvant efficacy evaluation and restaging holds significant clinical relevance and scientific value. It is crucial for standardizing the clinical diagnostic and treatment workflows of LARC, improving the treatment accuracy, and facilitating the establishment of individualized treatment regimens.

TIMING AND CRITERIA FOR EVALUATION AFTER NAT FOR RECTAL CANCER

Timing of the evaluation

The results of post-NAT evaluation serve as the basis for formulating subsequent treatment strategies. The rational determination of evaluation timing is important for improving the pathological complete response (pCR) rate and mitigating adverse outcomes caused by overly early or delayed intervention. Most of the current clinical guidelines and studies recommend restaging and evaluation 6-8 weeks following the completion of nCRT to achieve the optimal balance between “maximizing tumor regression” and “controlling disease progression and surgical risk”[6-9]. Multiple meta-analyses and systematic reviews have confirmed that conducting evaluations within this window significantly improves pCR rates without increasing surgical risks. Conversely, delaying evaluations beyond 14 weeks not only fail to further enhance pCR rates or diagnostic accuracy but may also increase the incidence of surgical complications due to progressive pelvic fibrosis[6,8,9]. However, in real-world clinical practice, the timing of evaluation still requires dynamic adjustment based on individual patient characteristics. For patients undergoing TNT or presenting high-risk clinical characteristics, the optimal evaluation strategy should be established after comprehensively considering the treatment response, underlying disease status, and limitations of imaging evaluations[10].

Evaluation criteria

Pathological evaluation remains the gold standard for assessing therapeutic efficacy and restaging following NAT in rectal cancer. Note that pCR is defined as the absence of residual viable tumor cells in the resected specimen (ypT0N0)[11]. Patients who achieve pCR typically demonstrate favorable long-term outcomes. It has been established as one of the most widely accepted and validated surrogate endpoints for assessing treatment response[12]. Nevertheless, a critical limitation of pathological evaluation lies in its inherent reliance on the availability of a completely resected specimen, which, however, is inconsistent with the concept of organ preservation. Clinical evaluation criteria play a pivotal role in the implementation of the “watch-and-wait” (WW) strategy. Clinical evaluation is commonly conducted by using a multimodal approach, which involves integrating digital rectal examination, endoscopic evaluation, and magnetic resonance imaging (MRI), among other modalities[13-15]. In 2021, Fokas et al[16] proposed a comprehensive set of criteria for defining the clinical complete response (cCR). A diagnosis of cCR requires that all of the following criteria must be simultaneously satisfied: (1) No palpable mass should be detected on digital rectal examination; (2) Endoscopic examination should not yield any visible residual tumor; only minor residual erythematous ulceration or scarring is allowed; and (3) High-resolution pelvic MRI should demonstrate substantial tumor regression, characterized by either the complete absence of any discernible residual mass or only a fibrotic replacement, alongside the absence of radiologically suspicious lymph nodes.

EVALUATION OF EFFICACY AFTER NAT FOR RECTAL CANCER

Imaging evaluation

Imaging restaging following nCRT is a pivotal step in comprehensive diagnostic workup and therapeutic decision-making in rectal cancer. Its accuracy is directly related to the selection of surgical strategy and the development of WW management protocols[17,18]. However, nCRT can induce complex histological changes, including tumor necrosis, tissue edema, and fibrosis[19], which significantly interfere with the interpretation of imaging signals, posing a challenge to the precise evaluation of lesions. In the subsequent sections, we discuss the advantages and limitations of four mainstream imaging techniques, namely computed tomography (CT), MRI, positron emission tomography (PET)/CT, and PET/MRI, in evaluating the treatment efficacy after nCRT. Table 1 lists the evaluation performance of different imaging, endoscopic, and other advanced techniques.

Table 1 Evaluation performance of different techniques in locally advanced rectal cancer patients after neoadjuvant chemoradiotherapy.

Evaluation methods	Area under the curve	Ref.
CT imaging histology	0.881	[25]
MRI, T2WI	0.88-0.91	[33]
MRI, gadolinium-enhanced	0.94	[38]
MRI-based tumor regression grade	0.729	[57]
Cell-free DNA	0.818 (0.886 when combined with MRI evaluation)	[57]
Deep learning algorithm	0.713	[63]
Endoscopy-based deep convolutional neural network	0.77	[64]
Contrastive language-image pretraining-based multimodal endorectal ultrasound	0.928	[65]
Gut microbial profile analysis	0.988	[72,78]

CT: Computed tomography; MRI: Magnetic resonance imaging; T2WI: T2-weighted imaging.

CT: CT is a widely used imaging tool for post-nCRT restaging. It is characterized by high accessibility and rapid examination speed. However, its accuracy in local T-staging and N-staging remains limited. Conventional contrast-enhanced CT typically performs restaging based on three key indicators: Tumor infiltration depth, lymph node morphology, and circumferential resection margin (CRM). Several domestic studies have reported the compliance rates of 63.6% and 52.3%, respectively, for T-staging and N-staging after nCRT in patients with low and intermediate rectal cancer. In contrast, previous international studies have reported even lower compliance rates[20,21]. The error in T-staging mainly stems from the overlapping imaging manifestations of the post-treatment scar tissue and residual tumor, while the error in N-staging is mostly attributed to the difficulty in distinguishing metastatic lymph nodes from the reactive ones.

Multislice CT has shown improvements over conventional CT in evaluating the depth of tumor infiltration. However, its accuracy in detecting lymph nodes is comparable to or slightly inferior to that achieved using MRI[20]. Notably, both multislice CT and MRI face challenges in accurately restaging patients with ypT0-1 stage rectal cancer[22], a finding that suggests the potential for appropriate simplification of the imaging evaluation process in certain patient subgroups. Despite the limited ability of CT in local staging, thoracic-abdominal-pelvic enhanced CT is uniquely valuable in the detection of distant metastases. This CT can identify new liver or lung metastases in approximately 10%-15% of patients, thereby altering the treatment paradigm[23].

In recent years, the emergence of CT imaging histology has provided a new direction for post-nCRT evaluation. Deep-learning-based imaging histology models can achieve an accuracy of 80% in pCR prediction[24]. The area under the curve (AUC) can be increased to 0.881 if CT imaging histology is combined with various clinical indicators[25]. Moreover, the accuracy of lymph node status prediction can reach 77.8%[26]. However, most of the current related studies are single-center, small-sample exploratory studies. Large-scale multicenter prospective studies are needed to further validate the reproducibility and stability of CT imaging histology[27].

MRI: MRI has become a core imaging tool for post-nCRT restaging of rectal cancer owing to its multisequence imaging capability and excellent soft tissue resolution. In the conventional sequence, T2-weighted imaging (T2WI) clearly delineates the tumor and intestinal wall hierarchy, while T1-weighted enhanced imaging (T1WI) is used to assess the tumor blood supply. The combination of T2WI and T1WI can achieve a diagnostic sensitivity of 81% and specificity of 67% for T3-T4 stage lesions[28]. However, for T0-T2 stage lesions, post-treatment necrosis, edema, and residual tumor often show overlapping manifestations in signal, resulting in a significant 25% decrease in MRI staging accuracy and susceptibility to overstaging[29]. Because underestimating and understaging cancer can lead to significantly poor outcomes, multimodality evaluations should be integrated to minimize the risk of understaging when adopting the WW strategy.

MRI-based evaluation of lymph nodes is conducted by considering the size, morphology, and enhancement features of the lymph nodes. However, the imaging presentation of metastatic and reactive lymph nodes is highly similar, resulting in diagnostic sensitivity and specificity of approximately 77%[28]. Arndt et al[30] reported that MRI staging results had limited concordance with pathologic staging and often showed a staging bias, which may be related to the persistent effect of NAT. Although mild overstaging usually does not affect subsequent treatment decisions, special care is needed in cases of cCR. In addition, staging underestimation exists in approximately 25% of cases, yet such patients often still benefit from NAT. Therefore, when evaluating early rectal cancer and developing a WW strategy, it is recommended to incorporate other complementary imaging techniques to improve diagnostic accuracy[30]. MRI has significant advantages over CT in the evaluation of anatomical structures. Its diagnostic sensitivity and specificity for the involvement of CRM were 85.4% and 80.0%, respectively[31,32].

In recent years, the development of imaging histology has further expanded the diagnostic potential of MRI. By extracting high-throughput features from multidimensional images such as T2WI and diffusion-weighted imaging (DWI), the pCR prediction ability can be significantly improved. The AUC of pCR predicted by a T2WI-based imaging histology model can reach 0.88-0.91[33], which can be enhanced to 0.97 if combined with DWI features[34], providing a solid basis for accurate identification of pCR and optimization of WW strategies.

The integration of functional MRI sequences has significantly improved the accuracy of restaging. Among these sequences, DWI quantifies water molecule motion through the apparent diffusion coefficient (ADC) to differentiate residual tumors with higher cell density from fibrotic scar tissues[35]. T2WI and DWI combination improves T staging compliance by approximately 10%[36]. In addition, a significant post-treatment increase in ADC values is considered to be an important imaging hallmark of pCR. Patients in the pCR group had significantly higher post-treatment ADC values than those in the non-pCR group[37].

Multiparametric imaging models based on histology have further optimized the diagnostic performance. Studies using advanced gadolinium-enhanced MRI techniques (e.g., fat-suppressed high-resolution three-dimensional-gradient-recalled echo-T1WI) have reported an AUC of up to 0.94 for the workup characteristic of a subject for the metrics related to enhancement homogeneity. The specificity for the benign-malignant differentiation of lymph nodes can exceed 85%[38]. However, Pangarkar et al[39] pointed out that the diagnostic accuracy of this technique is still insufficient for mucinous lymph nodes, suggesting the need for further improving tissue-specific modeling and multimodal fusion in the future.

PET: PET/CT. By integrating ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) metabolic imaging with CT anatomical imaging, PET/CT can achieve local restaging and distant metastasis screening in a single examination and has a unique advantage in evaluating the efficacy of treatment after nCRT. The accuracy rates afforded by PET/CT for T-staging and N-staging are 60% and 71%, respectively, which are overall comparable to those obtained using MRI. In addition, with an negative predictive value of up to 93.5%, PET/CT is significantly better than MRI in pCR prediction and provides an important reference for developing WW strategies[40,41].

PET/CT affords 97% accuracy in the detection of distant metastases. It is particularly efficient in identifying microscopic metastases that are easily missed by CT or MRI[40]. However, PET/CT has inherent limitations too. It has a lower spatial resolution, which affects the judgment of the depth of tumor infiltration. In addition, post-treatment radio inflammation can cause increased FDG uptake, which leads to false-positive results[40,42]. It is an expensive procedure that is also accompanied by a certain amount of radiation exposure, which restricts its application in preoperative routine restaging. Overall, for patients with rectal cancer undergoing nCRT, MRI remains a key tool for assessing T-staging and the lymph node status, while PET/CT is irreplaceable for pCR prediction and detecting distant metastasis[40].

PET/MRI. PET/MRI combines the advantages of FDG metabolic imaging with the high soft-tissue resolution of MRI, forming an excellent combination for post-nCRT restaging. PET/MRI outperforms MRI alone in localized tumor-node-metastasis staging: Crimì et al[43] showed that the sensitivity, specificity, and overall accuracy of T-staging ranged from 86% to 100%, from 40% to 94%, and from 77% to 98%, respectively, while the corresponding indexes for N-staging were 55%-100%, 75%-99%, and 77%-98%. This advantage mainly stems from the complementary effects of MRI’s precise localization at the structural level and PET’s metabolic signaling, which allow for a more reliable differentiation between residual tumor and scar tissue, thus significantly reducing the rate of N-staging misclassification. In terms of metastasis detection, PET/MRI and PET/CT exhibit similar sensitivities. Meta-analysis and comparative studies of non-small cell lung cancer have shown no significant differences in sensitivity or specificity between the two modalities for detecting lung metastases or distant metastases[44,45]. However, it has significant advantages in the detection of soft tissue metastases such as the liver: Its sensitivity in detecting liver metastases can reach up to 92%-98%, which is significantly higher than that obtained using PET/CT (66.7%-76.8%)[46-48]. PET/MRI is particularly advantageous for detecting small lesions (< 10 mm) that are often missed by PET/CT[46,48]. This advantage is also confirmed by relevant meta-analyses, with a pooled detection rate of 93%-96% for small lesions, compared with 77%-82% for small lesions obtained using PET/CT[49,50]. Although PET/MRI theoretically enables a comprehensive integration of morphology, function, and metabolism information, its clinical popularity is still limited by the scarcity of equipment, long examination time, and high cost. Currently, PET/MRI is primarily being used as a key supplement to PET/CT and demonstrates considerable potential in scientific research and the evaluation of precision radiotherapy programs.

Endoscopic evaluation

Colonoscopy: Colonoscopy is a key procedure for evaluating the therapeutic response and post-nCRT restaging in patients with LARC. It is particularly indispensable for determining whether a patient has achieved cCR and in guiding the implementation of the WW strategy. As the fundamental modality in post-nCRT response evaluation, colonoscopy or rigid sigmoidoscopy is primarily used to assess the presence of residual tumor(s) by observing features such as ulcer healing, tumor disappearance, and restoration of normal mucosal color and texture[16]. Notably, the absence of a visible residual tumor on endoscopic examination does not necessarily indicate the absence of tumor cells in the submucosa or deeper layers. This limitation contributes to a relatively high false-negative rate for colonoscopic evaluation, which may result in the underestimation of the residual disease in clinical practice. Therefore, when both imaging findings and endoscopic appearance support the diagnosis of cCR, biopsy may still yield false-negative results. Thus, the therapeutic response should not be judged solely based on biopsy outcomes. However, in cases where the diagnosis of cCR remains uncertain, biopsy serves as an important complementary evaluation tool. As demonstrated by Felder et al[51], biopsy can significantly enhance the diagnostic accuracy of endoscopically suspected cCR cases, a benefit that is particularly evident in non-experienced centers.

Endorectal ultrasound: Endorectal ultrasound (ERUS) is a conventional modality for the local evaluation of rectal cancer and is widely applied, particularly for T staging. The technique provides a high spatial resolution in assessing the depth of local tumor invasion within the rectal wall. However, its accuracy in post-nCRT restaging is significantly affected by tissue alterations, including fibrosis and inflammation induced by radiotherapy and chemotherapy[6]. Huang et al[52] conducted a study on 86 patients with rectal cancer who underwent NAT and reported an overall ERUS accuracy of only 67.4% for T restaging, further underscoring its limitations in post-treatment evaluation. In recent years, to enhance the restaging performance of ERUS, researchers have explored the integration of shear wave elastography (SWE), a novel imaging technique that quantitatively measures and analyzes tissue stiffness. SWE has shown promising advances in the evaluation of the therapeutic response among patients with LARC treated with nCRT. Cong et al[53] reported that the accuracy, sensitivity, and specificity of SWE for identifying T downstaging after nCRT were 89.0%, 87.7%, and 93.8%, respectively, markedly superior to the diagnostic efficacy of conventional ERUS. Similarly, Qian et al[54] incorporated SWE into the post-nCRT evaluation of LARC and found that its predictive accuracy for both T restaging and pCR significantly exceeded that obtained using traditional imaging modalities. These findings suggest that SWE represents a promising technological advancement for improving the precision of post-nCRT staging in rectal cancer.

Emerging technologies

Liquid biopsy: It is known that ctDNA, a fraction of cell-free DNA present in peripheral blood and other body fluids, represents a noninvasive biomarker that enables repeated and real-time monitoring of tumor burden and molecular evolution[55,56]. In contrast to traditional tissue biopsy, ctDNA testing offers clear advantages in terms of minimal invasiveness and dynamic surveillance of disease progression. Current research on ctDNA primarily focuses on its predictive value for pCR and cCR, as well as its potential for evaluating minimal residual disease and the recurrence risk. In a prospective study on 119 patients with LARC, Wang et al[57] reported that ctDNA clearance is significantly correlated with the pathological tumor regression grade (TRG). The AUC for ctDNA alone in predicting pCR was 0.818, which increased to 0.886 when combined with the MRI-based TRG, outperforming MRI-based TRG alone (AUC = 0.729). Lin et al[58], analyzing data from the UNION trial that investigated a short-course radiotherapy regimen combined with neoadjuvant chemotherapy and immunotherapy, found that ctDNA levels significantly declined during treatment and that ctDNA clearance was strongly associated with pCR. In the NOMINATE study, Akiyoshi et al[59] studied 64 patients with LARC and found that the baseline ctDNA positivity rate was as high as 98.4%. Among 25 patients who achieved cCR or near-cCR and underwent nonoperative management, the post-neoadjuvant ctDNA clearance rate reached 100%, compared with only 51% in 39 non-cCR patients. Further analysis revealed that ctDNA positivity at the time of post-NAT restaging had 100% specificity and a positive predictive value for detecting residual pathological disease and was significantly associated with shorter disease-free survival. Chang et al[60] conducted a systematic review and meta-analysis of 10 studies and reported similar results, that is, patients who remained ctDNA-positive after nCRT had significantly poorer recurrence-free survival and overall survival, as well as a lower likelihood of achieving pCR. Notably, postoperative ctDNA positivity demonstrated an even stronger predictive value for recurrence.

Multi-omics: Multi-omics approaches integrate information from multiple biological dimensions, including clinical features, radiomics, metabolomics, transcriptomics, and liquid biopsy, and provide a more comprehensive characterization of tumor heterogeneity compared with that achieved using single biomarkers and improve the accuracy of post-NAT response prediction in LARC[61]. Jiang et al[62] performed genomic sequencing of tumor tissue and plasma cell-free DNA from 16 patients with rectal cancer before and after NAT and conducted multi-omics integration by combining data with The Cancer Genome Atlas and proteomic databases. Their analysis demonstrated that copy number variation and tumor mutation burden effectively distinguish patients with a favorable vs poor response to neoadjuvant chemotherapy. Further investigation revealed that copy number variation gains in genes such as EGFR and HSP90AA1 were significantly associated with increased drug sensitivity and improved prognosis. Based on these findings, the study developed a treatment-related gene-based nCRT prediction score, which, when combined with tumor mutation burden, may serve as a potential tool for evaluating the therapeutic response and prognosis in patients with rectal cancer. The MOREOVER study[61] attempted to integrate multi-omics data, including MRI radiomics, dynamic ctDNA profiles, and gut microbiome composition, for enhancing the accuracy of pCR prediction models. In the study design, samples were collected at seven time points before, during, and after nCRT to construct a time-series-based integrated predictive algorithm for providing decision support for nonoperative management strategies.

AI: In recent years, AI, particularly deep learning algorithms, has been increasingly applied to colorectal endoscopic image analysis for predicting tumor response to NAT. Kato et al[63] trained a convolutional neural network on 403 colorectal endoscopic images and achieved a sensitivity of 77.6% and an AUC of 0.713 for predicting the treatment response. Thus, they successfully demonstrated the potential of AI-assisted evaluation in identifying patients with a poor response. Chen et al[64] included 1000 patients with LARC and developed a deep convolutional neural network based on endoscopic images. The model showed excellent performance in predicting pCR after NAT, achieving an AUC of 0.77 and a specificity of 92.98% in an independent test set, significantly outperforming experienced endoscopists. Zhang et al[65] introduced a contrastive language-image pretraining model for the analysis of ERUS images to predict TRG after nCRT. They retrospectively analyzed data from 577 patients with LARC, integrating ERUS images with patient clinical text information into the two-branch contrastive language-image pretraining architecture. The resulting AUCs were 0.928 and 0.900, respectively, both significantly higher than those achieved through conventional radiological evaluation. However, most of these AI-based studies are exploratory, single-center, and based on small sample sizes. Their generalizability, multi-institutional reproducibility, standardization, and cost-effectiveness should be thoroughly studied. In addition, clinicians should pay close attention to the regulatory approval status of AI-based technologies used in medical practice, which will help ensure that these cutting-edge innovations meet established standards for safety, effectiveness, and clinical reliability.

IMPACT AND CHALLENGES OF RESPONSE EVALUATION IN THE NEW ERA

TNT

The TNT approach has emerged as one of the standard treatment strategies for LARC, with the primary goal of achieving maximal early tumor regression and controlling the micrometastatic disease. However, TNT introduces new challenges for post-treatment restaging. First, the accuracy of imaging-based evaluation is limited. Studies have shown that, after TNT, the concordance between MRI evaluation and postoperative pathological findings declines. The sensitivity for detecting CRM positivity can be as low as 45%. In addition, up to 73% of pathologically positive lymph nodes are misclassified as negative on MRI; this discrepancy is particularly pronounced in low rectal cancers[66,67]. Second, the interpretation of imaging becomes more challenging. TNT-induced tissue changes, including fibrosis and radiation-associated injury, can mimic residual tumors on imaging or even obscure viable tumor foci, leading to both overestimation and underestimation of residual disease[6,66,67]. Third, the timing of evaluation is highly complex. TNT incorporates multiple regimens, such as induction and consolidation chemotherapy. Tumor regression exhibits substantial heterogeneity, making it difficult to determine a uniform, optimal time point for restaging. Notably, early evaluation may underestimate the clinical response, whereas delayed evaluation could postpone surgery or increase the risk of fibrotic complications. Consequently, there is currently no consensus on the optimal interval for restaging after TNT[6,67]. Finally, the existing evaluation criteria have several limitations. Traditional morphology-based response evaluation standards are often inadequate in the TNT setting, as some patients achieving pCR may not exhibit marked morphologic changes, making it challenging to accurately identify candidates suitable for the WW strategy.

Immunotherapy

In recent years, the emergence of immune checkpoint inhibitors and their application in NAT for rectal cancer have posed new challenges to the conventional post-treatment restaging system. Traditional restaging primarily relies on imaging and pathological evaluation of tumor downstaging. However, immunotherapy often induces atypical imaging responses and deep pathological remission, limiting the applicability of existing staging criteria for assessing therapeutic efficacy. Pseudoprogression, which is thought to result from immune cell infiltration, may occur during immunotherapy. In such cases, residual lesions on imaging may not correspond to viable tumor cells[68]. To address this challenge, the concept of immunotherapy RECIST has been proposed[69]. Under immunotherapy RECIST, when lesions enlarge or new lesions appear, they are not immediately classified as disease progression. Instead, they are recorded as having immune-unconfirmed progressive disease. In such cases, the follow-up imaging is performed 4-8 weeks later. Only if progression criteria are confirmed in this subsequent evaluation is the disease classified as true progression, thereby enhancing the accuracy of treatment response evaluation in the context of immunotherapy.

Intestinal flora

Intestinal flora as predictive biomarkers: In recent years, the composition and diversity of the intestinal flora have been recognized as key independent factors influencing the efficacy and toxicity response of NAT for LARC. A higher pretreatment flora α-diversity is significantly and positively associated with the achievement of pCR[70,71]. Compared to the overall microbial diversity, specific floral compositional features have a higher specificity as predictive biomarkers for pCR. Enrichment of butyrate-producing flora, represented by Faecalibacterium prausnitzii, has been confirmed to be closely associated with excellent treatment response and has become one of the most representative predictive microecological features[72-74].

Further macro-genomics studies have revealed that the high expression of specific functional gene pathways (e.g., short-chain fatty acid metabolism and immunomodulation-related genes) is closely associated with radiotherapy sensitivity and anti-inflammatory status, providing an important basis for the molecular mechanism of intestinal flora in the prediction of pCR[75,76]. In addition, altered flora structure is closely associated with radiotherapy toxicity. Some specific flora (e.g., an imbalance in the ratio of the phylum actinobacteria to the thick-walled phylum) can predict the risk of severe acute adverse reactions, such as radiation enteritis, thus potentially informing pretreatment risk stratification and individualized interventions[77].

AI prediction models based on gut microbial profiles are rapidly evolving. For example, the SPEED algorithm, which integrates macrogenomic features, and the random forest classification model perform well in pCR prediction, with an AUC as high as 0.988 that remained stable between 0.73 and 0.78 in an independent validation cohort[72,78]. Together, these results establish the biomarker value of intestinal flora in LARC precision medicine and provide a theoretical and technological basis for the construction of a novel prediction system.

Future prospects of flora analysis in clinical decision-making and restaging: Transforming colony features into actionable clinical tools is an important development direction in the current research on tumor radiotherapy. Several recent studies have attempted to combine colony features with machine learning algorithms to build high-precision models for predicting patient response to immunotherapy combined with radiotherapy regimens. Their predictive performance has demonstrated stability and reproducibility in multicenter validation[79]. Such models are expected to be an important tool for optimizing patient selection and may, in particular, provide more objective screening criteria for WW strategies, thereby reducing the risk of over- or undertreatment[80].

The future research trend lies in the deep integration of colony information with traditional imaging restaging techniques (e.g., MRI and PET/MRI) to construct a multidimensional comprehensive evaluation system covering metabolic, structural, and microecological features. Such interdisciplinary integrated models are expected to significantly improve the ability to differentiate post-treatment fibrosis from residual tumors[81-83], achieve more accurate restaging determination, and provide a solid basis for developing individualized treatment strategies for patients with LARC.

Organoids

Validation of organoid modeling and drug sensitivity prediction: As a promising research platform, PDOs have become an important tool for oncology research and precision medicine because of their ability to highly mimic the histological structure, genomic heterogeneity, and molecular features of primary tumors in vitro[84]. This technique is now widely used to study malignant tumors, including pancreatic cancer, biliary tract cancer, and brain tumors[85,86]. Its core strength lies in its ability to conduct in vitro drug sensitivity testing to predict the clinical response of a patient to a treatment regimen. A prospective blinded study in advanced colorectal cancer demonstrated that the drug sensitivity results of PDOs are highly consistent with the patients’ response to chemotherapy, fully validating their reliability in individualized efficacy prediction[62]. Similar results have been validated in solid tumors, such as bladder cancer and kidney cancer[87-89]. In addition, the high-throughput analysis strategy combining imaging and machine learning further improves the objectivity and efficiency of drug sensitivity screening[90].

Application of organoids in restaging and future challenges: The unique value of PDOs in the management of LARC lies in their ability to achieve dynamic and functional “in vitro restaging”. By constructing PDOs from biopsies and conducting drug sensitivity testing prior to NAT, it is possible to mimic the in vivo treatment response in vitro, thereby predicting the sensitivity of the tumor to the established regimen and providing clinicians with a functional reference for therapeutic decision-making[91]. This strategy provides important clues for treatment intensification or adjustment, driving the shift in restaging from morphological to functional evaluation. Despite the promising future, however, PDOs still face multiple challenges in routine clinical applications. Current research focuses on the construction of more complex three-dimensional co-culture systems, such as co-culturing with immune cells and fibroblasts to form “assemblage-like bodies”, in order to more realistically reproduce the tumor microenvironment[92] and more accurately assess the efficacy of NATs, including immunotherapy[93]. The introduction of emerging technologies such as bioprinting is facilitating the standardization and scale-up of PDO models[94]. However, poor culture success rates, long modeling, and drug sensitivity testing cycles, and the difficulty in achieving application within a limited clinical decision window remain key bottlenecks that limit its dissemination[95]. There is an urgent need to validate its clinical effectiveness and cost-efficiency through technology optimization and prospective clinical trials, so as to promote the real integration of PDOs into precision medicine practice.

CONCLUSION

Restaging following NAT for colorectal cancer is a critical step in formulating subsequent treatment strategies. Its accuracy directly influences the selection of surgical protocols and the implementation of the WW strategy. Although no consensus has yet been achieved regarding the optimal timing for restaging, most studies support performing imaging evaluation 6-8 weeks after the end of the treatment. It should be dynamically adjusted to take into account interpatient variability in clinical practice. Pathologic assessment remains the gold standard for determining treatment efficacy. However, its dependence on postoperative specimens contradicts the current concept of “organ preservation”.

At present, restaging is mainly performed by using various imaging techniques. MRI, with its excellent soft tissue resolution, is considered the gold standard for localized T staging, especially after the application of high-resolution T2-weighted sequences, which has significantly improved the evaluation accuracy. However, the sensitivity and specificity of the imaging techniques are still limited in determining the lymph node status. In addition, the overall diagnostic accuracy remains moderate, and the cost, accessibility, and diagnostic efficacy of different imaging techniques differ. Therefore, a comprehensive evaluation system based on multimodal information should be established instead of relying on a single imaging technique. MRI is recommended as the core tool for local staging, along with chest-abdomen-pelvis CT to screen for distant metastases. For patients with significant efficacy and a planned WW strategy, PET/CT or PET/MRI can be further used to improve the judgment accuracy.

Looking ahead, with the rapid development of functional imaging standardization, multimodal image fusion, image histology and AI technologies, as well as the continuous maturation of emerging biological tools such as intestinal flora and organoids, the evaluation of rectal cancer restaging is gradually moving toward a more precise, intelligent, and individualized paradigm. This trend will provide a more scientific and dynamic basis for making clinical treatment decisions. Given the current challenges of poor evaluation accuracy and reproducibility, future developments should focus on the following two core areas: The integration of multi-omics information data and the construction of an individualized dynamic evaluation model. Centered on patient-specific risk stratification, such models will incorporate early treatment response signals (e.g., dynamic imaging changes and biomarker fluctuations) for real-time optimization of the content and timing of the imaging evaluation program. The optimal follow-up intervention strategy, whether surgical resection or non-surgical curative plan, can be accurately matched to each rectal cancer patient receiving TNT during disease management by using this dynamic, closed-loop restaging system. This may promote the treatment of rectal cancer from “standardization” to “precision medicine” in the true sense of the word and ultimately lead to improved therapeutic efficacy and a better quality of life.