Computed tomography-dominant surveillance strategies for colorectal cancer: Improving early detection of recurrence

Abstract

Colorectal cancer (CRC) is one of the most prevalent cancers globally, with a high recurrence rate following curative surgery, especially within the first 3 to 5 years. Post-surgical follow-up plays a vital role in detecting local and distant recurrences, significantly influencing survival rates. However, despite established guidelines recommending surveillance strategies, discrepancies persist regarding the optimal surveillance modality and patient adherence to follow protocols. Sala-Miquel et al’s study emphasize the superiority of computed tomography in detecting metastasis and recurrence, while also shedding light on the critical role of adherence to surveillance protocols in improving patient outcomes. This editorial discusses the implications of these findings for clinical practice, providing a comprehensive overview of the current landscape of CRC surveillance and the path forward for improving patient outcomes.

Key Words: Colorectal cancer; Post-surgical surveillance; Recurrence detection; Computed tomography scans; Tumor markers

Core Tip: Early and effective surveillance is crucial for detecting recurrence in non-metastatic colorectal cancer (CRC) patients’ post-surgery. While computed tomography scans remain the most effective diagnostic tool, combining them with tumor markers and colonoscopy offers a more comprehensive approach. Adherence to follow-up protocols is strongly linked to improved survival outcomes, underlining the importance of consistent patient engagement in post-surgical care. Personalized follow-up strategies, including the integration of advanced biomarkers and digital health tools, hold potential for optimizing CRC surveillance and improving patient outcomes.

TO THE EDITOR

Colorectal cancer (CRC) remains a leading cause of cancer-related mortality worldwide, the latest epidemiological data show that in 2022, the number of new cases of CRC will reach 1.92 million (9.6% of malignant tumors) and the number of related deaths will exceed 900000 (9.3%), which will rank third in the global spectrum of cancer incidence and second in the spectrum of causes of death, respectively[1]. Despite advancements in surgical techniques and adjuvant therapies[2], recurrence poses a significant challenge, with up to 30% of patients with stage II to III tumors experiencing relapse within 3 to 5 years post-curative resection[3,4]. Effective post-surgical surveillance is crucial for early detection of recurrence, enabling timely intervention to improve survival outcomes before disease progression[5,6]. Indeed, guidelines from leading expert groups, including the American Society of Clinical Oncology[7,8], the European Society for Medical Oncology[9,10], the American Society of Gastrointestinal Endoscopy[11], the National Comprehensive Cancer Network[12], and The Chinese Society of Clinical Oncology[13], all recommend intensive postoperative surveillance for patients with resected stage II and III CRC who are considered candidates for aggressive treatment, such as surgery. However, current surveillance guidelines have notable limitations that hinder optimal care. Uniform protocols are often applied across diverse risk profiles, failing to differentiate between low-risk stage I disease (recurrence rate about 5%-10%) and high-risk stage III disease (recurrence rate about 30%-40%), leading to potential over- or under-surveillance[3,12]. Moreover, the integration of advanced biomarkers, such as circulating tumor DNA (ctDNA), remains limited, despite evidence of its prognostic value[14,15]. These gaps underscore the need for personalized, risk-adapted surveillance strategies that incorporate novel diagnostics and patient-specific factors to enhance early detection while addressing ethical and economic challenges, such as equitable access to advanced technologies. A recent study by Sala-Miquel et al[16] highlights the diagnostic yield of surveillance modalities and the impact of adherence to follow-up protocols, offering valuable insights into optimizing CRC management. This editorial discusses the implications of these findings, providing a comprehensive overview of current CRC surveillance practices and future directions for improving patient outcomes.

Diagnostic yield of CRC monitoring models: Insights from the Sala-Miquel et al’s study[16]

Sala-Miquel et al's retrospective cohort study[16], involving 574 patients who underwent curative resection for stage I-III CRC, provides valuable insights into the diagnostic performance of commonly used surveillance modalities: Colonoscopy, computed tomography (CT) scans, and tumor markers (TMs). The authors' findings robustly demonstrate the superior diagnostic yield of CT scans in detecting recurrences and metastases, identifying nearly 90% of the detected events. This observation aligns with the understanding that CRC recurrence often presents as distant metastasis, which is more readily identified through cross-sectional imaging. However, the retrospective design of the study introduces potential limitations that warrant careful consideration. Selection bias from non-random patient inclusion and variability in CT protocols (e.g., contrast use, slice thickness), which may skew recurrence detection rates. Additionally, variability in CT protocols, such as differences in contrast use, slice thickness, or imaging frequency, could influence the sensitivity and specificity of recurrence detection. For example, non-contrast CT scans or thicker slice intervals may miss small metastatic lesions, while standardized protocols with thin-slice, contrast-enhanced imaging could enhance detection but increase costs and radiation exposure. These factors limit the generalizability of the study’s findings and highlight the need for prospective studies with standardized imaging protocols to validate the diagnostic yield of CT scans. Furthermore, the study’s reliance on historical data may not fully capture the benefits of recent advancements in CT technology, such as iterative reconstruction and dual-energy CT, which could improve detection of subtle metastases. The study also highlights the limitations of relying solely on TMs, with a significant proportion of recurrences (approximately 30%) being missed when using TMs alone, particularly in cases of lung metastasis. However, it's important to acknowledge potential limitations inherent in retrospective studies. The data may be subject to selection bias, and the completeness of follow-up data can vary. Furthermore, the specific CT protocols used may not be standardized across all patients, potentially influencing the sensitivity of detection. Recent advancements in CT technology, such as iterative reconstruction and dual-energy CT, could further enhance the detection of subtle metastases, although their cost-effectiveness in routine surveillance needs further evaluation. To provide a concise overview of these modalities’ strengths, limitations, and future directions, Table 1 summarizes their advantages, disadvantages, and potential areas for improvement, offering practical guidance for optimizing CRC surveillance[17-24].

Table 1 Advantages, disadvantages, and potential improvements for colorectal cancer surveillance modalities.

Modality	Advantages	Disadvantages	Potential areas for improvement
CT scans	High sensitivity for detecting distant metastases (88.9% diagnostic yield)[16]. Non-invasive, widely available[6]. Effective for both local and systemic recurrence[17]	Radiation exposure, particularly cumulative risk in younger patients[6]. Variability in protocols (e.g., contrast use, slice thickness) affects sensitivity[16]	Implement low-dose CT protocols to reduce radiation while maintaining sensitivity[6]. Standardize imaging protocols (e.g., thin-slice, contrast-enhanced) for consistent detection[16]
Colonoscopy	Direct visualization and biopsy capability for local recurrence and metachronous neoplasms[16]. High specificity for colorectal lesions[12]. Allows therapeutic intervention (e.g., polyp removal)[11]	Invasive, requiring sedation and bowel preparation[12]. Low detection rate for distant metastases (4.6%)[16]. Patient discomfort and adherence barriers[18]	Enhance patient education and preparation to improve adherence[19]. Prioritize high-risk patients using ctDNA to reduce unnecessary procedures[20]. Explore less invasive alternatives (e.g., virtual colonoscopy)[6]
Tumor markers (CEA, CA19-9, ctDNA)	Non-invasive, repeatable testing[21]. ctDNA offers high sensitivity for minimal residual disease (HR: 17.5 for recurrence)[14]. Guides risk-stratified surveillance[20].	Low sensitivity for CEA/CA19-9 (about 40% of recurrences missed, especially lung)[21,22]. False positives (e.g., CEA elevations in non-malignant conditions)[23]	Improve CEA/CA19-9 specificity via machine learning (e.g., microRNA integration)[24]. Standardize ctDNA protocols and reduce costs through subsidized programs[20]. Validate ctDNA in diverse populations to minimize false positives[14]

CEA: Carcinoembryonic antigen; ctDNA: Circulating tumor DNA; HR: Hazard ratio; CA19-9: Carbohydrate antigen 19-9; CT: Computed tomography.

The evolving role of oversight: A multi-modal approach

As evidenced by Sala-Miquel et al[16], CT is the most effective tool for detecting CRC recurrences, including both local and distant metastases, with a diagnostic yield of 88.9%. While CT combined with colonoscopy and colonoscopy alone had detection rates of 6.5% and 4.6%, respectively. This finding underscores the centrality of CT in post-surgical surveillance and aligns with prior studies supporting its utility in the early detection of recurrence. For instance, in a study involving 1,050 consecutive patients who underwent surgery for CRC, of the 176 patients, CT triggered the detection of recurrence in 137 (78%)[17]. Although Frakulli et al’s study[25] focused on non-small cell lung cancer post-stereotactic body radiation therapy, it reported CT’s sensitivity of 84.6% and specificity of 71.8% for recurrence prediction, further underscoring CT’s diagnostic reliability across cancer types. However, Sala-Miquel et al[16] note that TMs and colonoscopy have significant limitations, particularly for detecting lung metastases, where TM elevation occurred in only 41% of cases. A study evaluating intratumoral metabolic heterogeneity indices derived from 18F-FDG positron emission tomography (PET)/CT demonstrated that the heterogeneity index, embolic status, and regional lymph node status were significant predictors of recurrence following radical surgery for stage II/III CRC. These factors outperformed other conventional parameters and 18F-FDG PET/CT heterogeneity indices[26]. This reinforces the need for a multi-modal surveillance approach, combining CT with other diagnostic tools, to maximize detection sensitivity.

In recent years, the question of whether intensive surveillance provides a survival benefit has been a topic of ongoing debate. Several studies have sought to compare intensive follow-up strategies, typically involving annual CT scans, colonoscopies, and TM testing, to less frequent testing regimens. For example, the FACS trial[27], found no significant difference in survival between intensive follow-up with CT and carcinoembryonic antigen (CEA) testing and minimal follow-up strategies. A retrospective cohort study[28] involving 2333 patients with non-metastatic CRC assessed body composition using CT scans preoperatively, at 6 months to 1 year, and at 2 to 3 years postoperatively. The study found significantly lower 5-year overall survival (OS) rates in patients with preoperative sarcopenia compared to those without [95.8% vs 92.1%, hazard ratio (HR) = 2.234, P < 0.001], and similarly, 5-year recurrence-free survival was also lower (93.2% vs 86.2%, HR = 2.251, P < 0.001). Both preoperative and postoperative sarcopenia, as well as changes in disease status during surveillance, were found to be associated with tumor prognosis. These discrepancies may be explained by differences in study populations, the timing of follow-up, and the surveillance modalities used.

Impact of treatment adherence on survival outcomes

One of the critical insights from the study by Sala-Miquel et al[16] is the significant impact of adherence to follow-up protocols on both CRC-specific and OS. The study found that patients who adhered to follow-up guidelines had a lower risk of mortality, suggesting that the early detection of recurrences through consistent surveillance is associated with improved patient outcomes. This finding is consistent with previous research indicating that adherence to surveillance is a strong predictor of survival[13,29]. One of the most intriguing aspects of a meta-analysis[30], which compiled data from eleven randomized controlled trials, totaling 4055 patients (2330 men and 1725 women) with stage I to III colon or rectal cancer who underwent curative surgery is the discrepancy between OS and disease-specific survival. Indeed, while intensive surveillance was associated with a 25% improvement in OS, no effect was observed on CRC-related mortality. Moreover, a systematic review and meta-analysis of five randomized trials[31], involving 1342 patients post-radical resection for CRC, evaluated the impact of intensive follow-up on survival. Intensive follow-up significantly reduced all-cause mortality (pooled risk ratio: 0.81, 95%CI: 0.70–0.94, P = 0.007). This effect was most notable in four trials employing CT and frequent serum CEA measurements (risk ratio: 0.73, 95%CI: 0.60–0.89, P = 0.002). Intensive follow-up also enabled earlier detection of recurrences (mean difference: 8.5 months, 95%CI: 7.6–9.4 months, P < 0.001) and increased identification of isolated localized recurrences (risk ratio: 1.61, 95%CI: 1.12–2.32, P = 0.011). Furthermore, Mokhles et al[32] searched MEDLINE (Ovid), Embase, the Cochrane Library, Web of Science, and other databases for randomized trials comparing intensive surveillance with standard follow-up after primary CRC resection. Sixteen trials were included, with 11 reporting survival outcomes. Intensive surveillance advanced diagnostic testing for recurrence detection by a median of 10 months. However, 10 of the 11 studies found no significant difference in OS.

However, adherence rates in the Sala-Miquel et al[16] study were alarmingly low, with a steady decline observed over the follow-up period. The factors influencing adherence included older age, stage III disease, and the type of surgery performed. Beyond these patient-specific factors, systemic barriers likely contribute significantly to the observed decline in adherence. Limited access to healthcare facilities, particularly in rural or underserved areas, can prevent patients from attending follow-up appointments. Socioeconomic disparities, such as financial constraints or lack of transportation, may further exacerbate non-adherence, as patients may prioritize immediate needs over long-term surveillance. Cultural attitudes toward healthcare and varying levels of health literacy can also influence engagement with follow-up protocols. These systemic issues were not fully explored in Sala-Miquel et al’s study[16], which limits the understanding of adherence challenges in diverse populations. Future research should investigate these barriers through qualitative and quantitative approaches to identify targeted interventions, such as community-based outreach programs, subsidized transportation, or telehealth services, to improve adherence rates. Addressing these systemic factors is critical to ensuring equitable access to surveillance and maximizing the survival benefits of early recurrence detection. The decline in adherence over time mirrors findings from other studies, where adherence rates for CRC surveillance generally decrease after the first few years[18]. The challenge of maintaining long-term follow-up engagement underscores the need for innovative solutions to improve patient adherence. Personalized follow-up plans, enhanced patient education, and the use of digital health tools may play crucial roles in improving adherence rates and, ultimately, survival outcomes.

TM: A supporting role in surveillance

TM play a critical role in CRC surveillance but are insufficient as standalone tools due to variable sensitivity. Traditional markers, such as CEA and carbohydrate antigen 19-9 (CA19-9), are widely used but have significant limitations. The Sala-Miquel et al[16] study found that relying solely on CEA and CA19-9 missed approximately 40% of recurrences, particularly lung metastases, as these markers are more effective for detecting liver metastases[21,22]. Furthermore, a recent study reported that false-positive CEA elevations, ranging from 5 to 15 ng/mL, were frequently observed in patients monitored for disease recurrence following primary treatment for stages I-III CRC[23]. This underscores the need for complementary diagnostic modalities, such as CT scans, to achieve comprehensive surveillance.

The recent development of novel biomarkers, particularly ctDNA, offers significant advancements in recurrence detection. ctDNA detects minimal residual disease by identifying tumor-specific DNA fragments in the bloodstream, providing higher sensitivity than traditional markers[14,15]. For instance, a study involving 299 patients with stage I to III CRC evaluated for ctDNA found that, at postoperative month 1, ctDNA-positive patients were 17.5 times more likely to experience recurrence than ctDNA-negative patients (HR = 17.5; 95%CI: 8.9-34.4; P < 0.001), with a median lead time of 3.3 months over radiological methods (interquartile range: 0.5-6.5 months)[14]. The DYNAMIC-III trial, a 2023 randomized study, further demonstrated that ctDNA-guided surveillance in stage III colon cancer improved OS compared to standard care by enabling earlier detection and tailored adjuvant therapy[33]. This trial highlights ctDNA’s potential to optimize surveillance by prioritizing high-risk patients for intensive monitoring. Moreover, the 2023 update to the European Society for Medical Oncology guidelines endorses ctDNA as a tool for recurrence monitoring in stage II-III CRC, recommending its integration into surveillance protocols to guide imaging frequency[20]. This early detection capability enables timely interventions, potentially improving survival outcomes. Moreover, ctDNA can guide risk-stratified surveillance protocols, where ctDNA-positive patients are prioritized for more frequent imaging or colonoscopies, optimizing resource allocation. However, ctDNA also presents challenges. Its high sensitivity can lead to false-positive results, particularly in patients with non-malignant conditions that elevate circulating DNA levels, necessitating confirmatory imaging or biopsies. Moreover, the cost of ctDNA testing, which can range from hundreds to thousands of dollars per test, and its limited availability in resource-constrained settings, restrict its widespread adoption. Further validation is needed to standardize ctDNA protocols and assess cost-effectiveness for routine use. Despite these limitations, combining ctDNA with CT scans could enhance the sensitivity and specificity of surveillance protocols, offering a more comprehensive approach to recurrence detection.

Other novel biomarkers, such as fecal extracellular vesicle microRNAs, also show promise. A machine learning-assisted study demonstrated a diagnostic accuracy of 97.4% in 38 CRC patients, significantly outperforming CEA (26.3%) and CA19-9 (7.9%)[24]. While these biomarkers enhance detection, their integration with imaging remains essential. The high cost, need for specialized equipment, and limited evidence on long-term outcomes necessitate further research to establish their role in clinical practice. In summary, while traditional and novel biomarkers like ctDNA and microRNAs improve recurrence detection, their limitations—cost, availability, and false positives—highlight the need for a multi-modal approach combining biomarkers with CT scans to maximize diagnostic yield and inform surveillance strategies.

Personalized follow-up strategy: The way forward

Personalized follow-up strategies, tailored to individual patient risk profiles, are increasingly recognized as critical for optimizing post-surgical CRC care. By integrating clinical factors, molecular profiling, and advanced technologies, clinicians can design surveillance protocols that balance early recurrence detection with risks such as radiation exposure, particularly for younger patients. These strategies aim to enhance outcomes while minimizing unnecessary testing and improving patient adherence.

Risk stratification is central to personalized surveillance. Clinical factors, such as tumor stage (e.g., stage III vs stage I), age, and surgical type, alongside molecular profiling, guide surveillance intensity. For instance, microsatellite instability-high (MSI-H) tumors, present in approximately 15% of CRC cases, are associated with better prognosis and lower recurrence risk in stage II disease, potentially justifying less frequent imaging compared to microsatellite stable (MSS) tumors, which have higher recurrence rates[12]. The National Comprehensive Cancer Network (NCCN) guidelines recommend considering MSI status when planning surveillance, as MSI-H patients may benefit from extended intervals between CT scans[12].

Recent advancements in artificial intelligence (AI) offer promising tools to enhance personalized surveillance and address adherence challenges. AI-driven predictive models can stratify patients based on risk by analyzing clinical and laboratory data. For instance, Fan et al[34] utilized preoperative portal-phase CT scans to extract radiomics features, employing the LASSO method to develop a radiomics score (Rad-score). Their radiomics-based model accurately predicted recurrence risk in patients with stage II CRC. Nomograms integrating the Rad-score with clinical variables outperformed clinical-only models, achieving area under the curve values of 0.954 and 0.906 in the training and validation cohorts, respectively, compared to 0.765 and 0.705 for clinical models. This approach may enhance prognostic accuracy beyond traditional clinicopathologic risk factors, aiding in the identification of patients likely to benefit from adjuvant therapy. A study by Waljee et al[35] developed a Multianalyte Assay with Algorithmic Analysis using longitudinal complete blood counts, comprehensive metabolic panels, and patient demographics to predict gastrointestinal tract cancers, including CRC, with an area under the receiver operator curve of 0.750 (95%CI: 0.729-0.771). Adapting such models to postoperative settings could stratify recurrence risk by integrating clinical data, enabling more frequent monitoring for high-risk patients and less intensive schedules for others. Additionally, AI-powered histopathological tools, such as the VIPR algorithm described by Waljee et al[35], enable high-throughput analysis of hematoxylin and eosin-stained slides to detect subtle malignant features, potentially improving the identification of residual disease in post-surgical settings. However, these technologies require large, diverse datasets for validation, and their computational infrastructure can be costly, particularly in low-resource settings like sub-Saharan Africa, where genetic and environmental factors may affect model performance. False-positive predictions also risk unnecessary testing, underscoring the need for rigorous validation.

Balancing surveillance intensity with radiation risks is critical, especially for younger patients who face long-term risks from cumulative radiation exposure. For high-risk patients (e.g., stage III, MSS tumors), annual CT scans may be justified for the first 3-5 years, as recommended by NCCN guidelines[12]. However, for younger patients or those at lower risk (e.g., stage I, MSI-H), low-dose CT protocols or alternative imaging, such as magnetic resonance imaging (MRI), can reduce radiation exposure while maintaining diagnostic yield. A study by Ladabaum et al[6] suggests that low-dose CT protocols achieve comparable sensitivity for metastasis detection with significantly reduced radiation doses, making them suitable for younger patients.

The 2023 European Society for Medical Oncology guidelines further support the use of low-dose CT and MRI in CRC surveillance to minimize radiation risks, particularly for patients under 50, while ensuring timely detection of recurrences[20]. Clinicians should weigh these options, prioritizing low-dose CT or MRI for patients under 50 to minimize long-term risks, while ensuring high-risk patients receive timely imaging.

Adherence to follow-up protocols remains a challenge, with Sala-Miquel et al[16] reporting low adherence rates linked to patient factors (e.g., older age, stage III disease) and systemic barriers (e.g., limited healthcare access, socioeconomic disparities, transportation issues). These challenges, particularly in rural or underserved areas, necessitate targeted, evidence-based interventions to ensure equitable surveillance. Nurse-led follow-up programs enhance adherence through personalized education and regular follow-up calls. A randomized controlled trial by Lin et al[36] revealed that nurse - led interventions led to significant improvements in multiple aspects. These included the quality of discharge teaching, patients' readiness for hospital discharge, stoma self efficacy, the quality of life related to stomas, a reduction in complications, and a decrease in unplanned readmission rates. Telehealth platforms, enabling remote symptom and biomarker monitoring, can reduce access barriers. A randomized controlled trial by Lau et al[19] found that digital interventions increased CRC screening adherence by 15% in average-risk populations, suggesting similar potential for postoperative surveillance. AI-powered mobile applications with automated reminders and symptom-tracking features can further enhance engagement. However, implementing these solutions requires addressing digital literacy gaps and infrastructure limitations, especially in low-resource settings. Future research should explore community-based outreach, subsidized transportation, and telehealth integration to overcome systemic barriers and improve adherence. By integrating these technologies into clinical practice, healthcare providers can offer more accessible, patient-centered care, which may lead to better outcomes and enhanced patient satisfaction, and pave the way for more effective post-surgical care, particularly for high-risk populations. To operationalize personalized surveillance, we propose the following risk-adapted algorithm for stage I-III CRC patients post-surgery, based on clinical, molecular, and biomarker data (Table 2).

Table 2 Risk-adapted surveillance strategy for postoperative stage I-III colorectal cancer patients.

Risk category	Imaging (CT/MRI)	Colonoscopy	Biomarker testing (CEA)	Rationale
High-risk (e.g., stage III, MSS, positive biomarkers)	Every 6 months for years 1-3, annually for years 4-5 (low-dose protocols for patients < 50)	At 1 year post-surgery, then every 3 years if negative	Every 3 months for 3 years, then every 6 months	Higher recurrence risk justifies intensive imaging, balanced with low-dose CT to reduce radiation exposure[6,10,12]
Intermediate-Risk (e.g., stage II, MSS, negative biomarkers)	Annually for years 1-3, every 2 years for years 4-5 (MRI preferred for patients < 50)	At 1 year, then every 3-5 years if negative	Every 6 months for 3 years, then annually	Moderate risk warrants regular but less frequent imaging, with MRI to minimize radiation in younger patients[10,12]
Low-risk (e.g., stage I, MSI-H, negative biomarkers)	Every 2 years for years 1-5 (MRI or low-dose CT for patients < 50)	At 1 year, then every 5 years if negative	Annually for 5 years	Lower recurrence risk supports extended intervals, prioritizing radiation reduction[6,10,12]

MSS: Microsatellite stable; MSI-H: Microsatellite instability-high; CEA: Carcinoembryonic antigen; MRI: Magnetic resonance imaging; CT: Computed tomography.

CONCLUSION

In conclusion, the study by Sala-Miquel et al[16] reinforces the importance of a multi-modal surveillance strategy in optimizing post-surgical follow-up for non-metastatic CRC. CT scans remain the most effective tool for detecting recurrences and metastases, while TM and colonoscopy should be considered as supplementary methods. Adherence to follow-up protocols is a key determinant of survival, highlighting the need for strategies to improve patient engagement over the long term. As we move forward, the integration of personalized follow-up strategies and the incorporation of emerging biomarkers, such as ctDNA, AI could revolutionize CRC surveillance, allowing for more tailored, efficient, and effective care. By continuously refining follow-up protocols and embracing innovative solutions, healthcare providers can enhance early detection, improve survival outcomes, and ultimately provide better quality of life for CRC patients.