Skip to main content
NCBI home page
Annals of Gastroenterological Surgery logoLink to Annals of Gastroenterological Surgery
. 2026 Mar 24;10(5):1450–1462. doi: 10.1002/ags3.70218

Diagnostic Accuracy of Size‐Based Preoperative CT Assessment for Predicting Lymph Node Metastasis in Colon Cancer: A Systematic Review and Meta‐Analysis

Yuji Takayama 1,✉, Tomoya Shimizu 1, Jun Watanabe 2,3,4, Shimpei Ogawa 5, Yukihide Kanemitsu 6
PMCID: PMC13393513  PMID: 42495671

ABSTRACT

Background

Accurate identification of preoperative lymph node metastasis is essential for planning colon cancer treatment. Computed tomography (CT) is widely used for staging, but its diagnostic performance based on size criteria alone remains unclear. This study aimed to evaluate the diagnostic accuracy of preoperative CT for detecting lymph node metastasis in colon cancer.

Methods

A systematic search of MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials was conducted on June 12, 2025. Studies comparing preoperative CT with pathological evaluation and providing 2 × 2 contingency tables were included. Pooled sensitivity and specificity were calculated using a hierarchical summary receiver operating characteristic model and a bivariate random‐effects model. Subgroup analyses were performed according to cancer location (colon only vs. colon plus rectal cancer) and diagnostic criteria (size alone vs. size plus morphology). Study quality was assessed using QUADAS‐2.

Results

Twenty‐nine studies involving 5634 patients were included. The pooled sensitivity and specificity of CT for detecting lymph node metastasis were 0.693 (95% CI: 0.636–0.744) and 0.660 (95% CI: 0.581–0.731), respectively, with an area under the curve of 0.727. Meta‐regression showed no statistically significant differences in diagnostic performance between colon‐only studies and those including rectal cancer (p = 0.561 and 0.316), or between size‐only and morphologic criteria (p = 0.822 and 0.536).

Conclusions

CT using size criteria alone for preoperative lymph node staging in colon cancer demonstrated only moderate diagnostic performance. These findings indicate that relying on lymph node size as the primary determinant may be insufficient for reliable clinical decision‐making.

Keywords: colon neoplasms, lymphatic metastasis, meta‐analysis, systematic review, tomography, X‐ray computed

Preoperative CT based on lymph node size shows moderate accuracy for detecting nodal metastasis in colon cancer. In this meta‐analysis of 29 studies (5,634 patients), pooled sensitivity and specificity were 0.69 and 0.66. Size‐based CT alone has limited value for clinical decision‐making.

graphic file with name AGS3-10-1450-g006.webp

1. Introduction

Accurate preoperative diagnosis of lymph node metastasis (LNM) in colon cancer (CC) is a critical issue in determining appropriate treatment strategies. Numerous studies have reported that LNM is a risk factor for recurrence and is associated with poor prognosis [1, 2, 3, 4]. Furthermore, lymph node status is a key component of preoperative staging, influencing decisions regarding the extent of lymph node dissection [5] and the use of neoadjuvant treatment [6, 7]. Therefore, precise evaluation of LNM is of paramount importance in planning therapeutic strategies.

In rectal cancer, the role of preoperative magnetic resonance imaging (MRI) has been well established and incorporated into clinical guidelines [5, 8, 9]. In contrast, the clinical role of preoperative LNM assessment in CC is more nuanced. While colon resection and lymph node dissection are routinely performed, the recommended extent of lymph node dissection may vary depending on the preoperative staging classification. Current clinical guidelines [5] generally support D3 lymph node dissection when preoperative computed tomography (CT) suggests LNM. Conversely, D2 dissection is considered for patients without suspected LNM, particularly in early stages. Thus, preoperative LNM assessment can influence not only oncological risk stratification but also the planned extent of lymph node dissection. In a meta‐analysis by Nerad et al. [10], the sensitivity and specificity of CT for detecting LNM were reported to be 71% and 67%, respectively; however, these data were derived from studies published before 2015.

By analyzing more recent evidence, it may be possible to better assess the advantages and limitations of CT in the staging of CC and to provide updated reference values useful for both daily clinical practice and research. This systematic review and meta‐analysis aimed to evaluate the diagnostic performance of preoperative CT in detecting LNM, with a particular focus on size‐based diagnostic criteria.

2. Methods

2.1. Protocol and Registration

This systematic review and meta‐analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses of Diagnostic Test Accuracy Studies (PRISMA‐DTA) guidelines [11]. The study protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD420251177701).

2.2. Eligibility Criteria

Studies examining LNM on preoperative CT and pathological findings, where preoperative CT included size measurements, were included in this review. Cases undergoing preoperative treatment, case studies, reviews, studies not written in English, and CT scans performed under nonstandard or non‐clinical conditions were excluded. Additional articles were identified by cross‐checking the references of the retrieved studies. Only studies in which the results were presented in a 2 × 2 contingency table, or from which an equivalent table could be extracted, were included. In studies that included both colon and rectal cancers, only the data on CC were extracted if they could be separated from those on rectal cancer. If separation was not possible, the item “Colon, Rectum” was entered in the “Location” column in (Table 1).

TABLE 1.

Characteristics of studies included in the systematic review.

No Author Year Design CT modality and detector‐row configuration Affiliation and number of interpreters Patients Age Gender (M/F) Location Cut‐off value
1 Balthazar 1987 Retrospective CT NA 90 34–92 (mean 61) 47/43 Colon Max diameter ≥ 10 mm or clusters of ≥ 3
2 Okizuka 1995 Prospective ultrafast CT NA 2 28 44–86 (mean 65) 15/13 Colon, Rectum Max diameter > 8 mm
3 Hundt 1999 Prospective CT Radiologist 2 37 42–78 (mean 55) 21/16 Colon, Rectum Max diameter ≥ 10 mm or clusters of ≥ 3 and 100 HU
4 Akiyoshi 2009 Retrospective MDCT, 16 Radiologist 1, Surgeon 1 65 37–84 (mean 62) 36/29 Colon, Rectum Max diameter > 5 mm
5 Uchiyama 2012 Prospective MDCT, 16 Radiologist 2 77 29–91 (mean 68) 55/22 Colon Max diameter > 5 mm
6 Duman 2012 Prospective MDCT, 64 Radiologist 1 73 30–83 (mean 57.8) 34/39 Colon, Rectum Max diameter ≥ 5 mm or clusters of ≥ 3
7 Lao 2013 Retrospective MDCT, 64 Radiologist 1 152 66 (median) 82/70 Colon Max diameter ≥ 10 mm or clusters of ≥ 3
8 Lim 2013 Prospective MDCT, NA Radiologist 2 136 76 (median) 57/79 Right Colon Max diameter ≥ 10 mm or enhancement
9 Vries 2014 Retrospective MDCT, 4 Radiologist 2 106 29–87 (median 70.5) 50/56 Colon Max diameter > 10 mm or clusters of ≥ 3
10 Nørgaard 2014 Retrospective MDCT, 64 Radiologist 1 74 32–90 (mean 72) 29/45 Colon Max diameter > 5 mm
11 Elibol 2015 Retrospective MDCT, 16, 64 Radiologist 2 194 24–92 (mean 66.7) 83/58 Colon Max short diameter ≥ 5 mm
12 Choi 2015 Retrospective CT, NA Radiologist 1, Surgeon 1 64 67.6 (mean) 26/38 Colon Max diameter > 10 mm or 7–10 mm + morphology
13 Wiegering 2015 Retrospective MDCT, 16, 64 Radiologist 2 210 70 (median) NA Colon Max short diameter ≥ 5 mm
14 Huang 2016 Retrospective MDCT, 64 Radiologist 2 326 14–90 (mean 61.2) 213/113 Colon, Rectum Max diameter > 10 mm or clusters of ≥ 3
15 Erik 2017 Retrospective MDCT, 64 Radiologist 1 119 32–91 (median 69) 56/63 Colon Max short diameter ≥ 5 mm
16 Bai 2017 Prospective MDCT, 64 Radiologist 2 76 32–74 (mean 54) 40/36 Colon, Rectum Max diameter > 10 mm
17 Sasaki 2017 Retrospective MDCT, 16 NA 2 370 24–93 (median 69) 240/130 Colon Max short diameter > 10 mm
18 Lee 2017 Retrospective MDCT, 64 Radiologist 1 220 63.8 (mean) 129/91 Colon, Rectum Max diameter ≥ 10 mm
19 Kumamoto 2019 Retrospective MDCT, 64 Radiologist 2 70 39–95 (median 70) 41/29 Colon, Rectum Max long diameter ≥ 7.05 mm
20 Liu 2019 Retrospective MDCT, 128, 320 Radiologist 2 116 36–83 (mean 60.9) 78/38 Left Colon Max diameter ≥ 7 mm or morphology
21 Miao 2020 Retrospective MDCT, 40 Radiologist 2 284 17–91 (median 61.7) 132/152 Colon Max long diameter ≥ 10 mm
22 Eresen 2020 Retrospective MDCT, 64 Radiologist 2 312 63.8 (mean) 165/147 Colon Max diameter > 10 mm
23 Mou 2021 Retrospective MDCT, 64 Radiologist 2 106 62.4 (mean) 57/49 Colon Max long diameter > 10 mm or < 10 mm + morphology
24 Kim 2022 Retrospective MDCT, 40, 64 Radiologist 1 507 63.0, mean 243/264 Right Colon Max short diameter ≧ 5 mm
25 Yukimoto 2022 Retrospective MDCT, 64 ≦ NA 1 541 23–92 (median 67) 314/227 Colon, Rectum Max short diameter ≧ 7 mm
26 Emoto 2023 Retrospective MDCT, 80, 256, 320 Radiologist 1, Surgeon 1 250 17–93 (median 68) 128/122 Colon, Rectum Max short diameter ≧ 5 mm
27 Engel 2024 Retrospective MDCT, 64 Radiologist 2 229 NA NA Colon Max diameter ≧ 10 mm or 7‐9 mm + morphology
28 Kawazoe 2025 Retrospective CT, NA Surgeon 2 694 68.2, mean 372/322 Colon Max short diameter ≧ 5.5 mm
29 Leonhardi 2025 Retrospective MDCT, 128 Radiologist 2 108 70.1, mean 59/49 Colon Max short diameter ≧ 10 mm
Open in a new tab

Abbreviations: CT, computed tomography; MDCT, multidetector computed tomography; NA, not available.

2.3. Literature Search

The literature search was conducted on June 12, 2025 in the following databases: MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL; Cochrane Library). The search terms used included “colon cancer” or “colorectal cancer”, “lymph node metastasis”, “computed tomography” and related terms (supplementary).

2.4. Study Selection and Data Collection

Two authors (YT, TS) independently screened the titles and abstracts of studies identified through the literature search. Duplicates were excluded by checking study titles and author names. Full texts of studies potentially meeting inclusion criteria were obtained for full‐text review. Reference lists of included studies were screened to identify additional relevant studies. Disagreements between reviewers regarding study assessment were resolved through discussion and the judgment of a third author (JW). All data extraction was performed by the same author and verified for accuracy.

2.5. Quality Assessment

The QUADAS‐2 tool [12] was used to assess the quality of included studies. Study quality was evaluated based on two categories: “risk of bias” and “applicability of concerns”. The risk of bias category comprised four domains: patient selection, index test, reference standard, and flow and timing. The concerns about applicability category comprised three domains: patient selection, index test, reference standard. Risk within each domain was assessed as “high”, “low”, or “unclear”. Modification to the QUADAS‐2 signaling questions was made to enhance methodological consistency with the review objectives. These adaptations did not alter the domain structure or risk‐of‐bias judgment process.

2.6. Subgroup and Sensitivity Analyses

To explore potential sources of heterogeneity in diagnostic performance, subgroup analyses were conducted according to two clinically relevant factors. These included the anatomical focus of the study population (CC alone versus colorectal cancer (CRC) including rectal involvement) and the type of CT‐based assessment criteria (size‐only versus size combined with morphological evaluation). For each subgroup, bivariate random‐effects models and hierarchical summary ROC (HS‐ROC) curves were generated, and comparative performance was evaluated through meta‐regression. Additionally, sensitivity analyses were performed to assess the robustness of the main findings. These included restricting analyses to studies using ≥ 64‐slice multidetector CT (MDCT) and, separately, to studies in which CT interpretation was performed by radiologists.

2.7. Statistical Analysis

Estimates of sensitivity, specificity, and diagnostic odds ratio (DOR) were extracted from each study and presented in forest plots. The HS‐ROC model was used to construct a summary ROC curve and to calculate pooled estimates of sensitivity, specificity, and DOR [13]. This model accounts for heterogeneity among studies and is considered a random‐effects model. A bivariate random‐effects meta‐analysis (Reitsma model [14]) was applied to simultaneously pool sensitivity and specificity while accounting for their intrinsic correlation and between‐study variability. Furthermore, for subgroup analysis, a bivariate meta‐regression model with tumor site as a covariate was constructed to investigate whether diagnostic performance differed between the two groups. Differences in sensitivity and specificity were evaluated based on the regression coefficients, and statistical significance was assessed using Wald tests. When the estimated relative effect was imprecise, interpretation was based on the confidence interval of the regression coefficient and its directionality rather than the point estimate. Publication bias was assessed using Deeks' funnel plot asymmetry test [15]. Between‐study heterogeneity was evaluated using I2 statistics, τ [2], τ, and H indices. A p‐value of less than 0.10 for the slope coefficient was considered to indicate statistically significant asymmetry.

All analyses were performed using the mada package (version 0.5.12) within R software (version 4.5.2, R Foundation for Statistical Computing, Vienna, Austria).

3. Results

3.1. Study Selection and Characteristics

The PRISMA flow diagram is shown in Figure 1. A literature search identified 3232 studies. After excluding 883 duplicates, 2349 studies were screened based on their titles and abstracts. Subsequently, 106 full‐text studies were reviewed to assess eligibility for inclusion in this review, of which 77 were excluded. Ultimately, 29 studies were included in this review (Table 1) [16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44]. The studies period spanned from 1987 to 2025, and 24 studies were retrospective. Twenty studies included only CC, while 10 studies included rectal cancer that could not be separated from CC. Eighteen studies reported size alone, and 11 studies included morphological evaluation. Studies that included three or more clustered lymph nodes in their positive criteria were classified as morphological evaluations.

FIGURE 1.

FIGURE 1

Open in a new tab

Flow diagram for study selection for inclusion in this review.

3.2. Quality Assessment

Table 2 summarizes the methodological quality of the included studies. In the patient selection domain, two studies were judged to have an unclear risk of bias. In the index test domain, thirteen studies were judged to have an unclear risk. In the reference standard domain, two studies were judged to have an unclear risk. In the flow and timing domains, all studies were judged to be low risk. Regarding the applicability category, in the patient selection domain, two studies were judged to have an unclear risk. In the index test and reference standard domain, all studies were judged to be low risk.

TABLE 2.

Methodological quality and applicability assessment using the QUADAS‐2 tool.

Author Year Risk of bias Concerns about applicability
Patient selection Index test Reference standard Flow and timing Patient selection Index test Reference standard
Emil J. Balthazar 1987 Low Low Low ? Low Low Low
Hiromi Okizuka 1995 Low ? Low ? Low Low Low
W Hundt 1999 ? Low Low Low Low Low Low
T. Akiyoshi 2009 Low Low Low Low Low Low Low
Shuichiro Uchiyama 2012 Low ? Low Low Low Low Low
Mustafa Duman 2012 Low ? Low Low Low Low Low
I‐ H. lao 2013 Low Low Low Low Low Low Low
M. Lim 2013 Low Low Low ? Low Low Low
F.E.E. de Vries 2014 Low Low Low Low Low Low Low
Anne Nørgaard 2014 Low ? Low Low Low Low Low
Funda Dinç Elibol 2015 Low Low Low Low Low Low Low
Audrey H Choi 2015 Low Low Low Low Low Low Low
Armin Wiegering 2015 Low Low Low Low Low Low Low
Yan‐qi Huang 2016 Low ? ? Low Low Low Low
Rollvén Erik 2017 Low Low Low Low Low Low Low
Rong‐jie Bai 2017 Low Low Low Low Low Low Low
Kazuhito Sasaki 2017 Low ? Low Low Low Low Low
Joo Young Lee 2017 Low ? Low Low Low Low Low
Tsutomu Kumamoto 2019 Low High Low Low Low Low Low
Li‐Heng Liu 2019 Low Low Low Low ? Low Low
Shi‐Song Miao 2020 Low ? Low Low Low Low Low
Aydin Eresen 2020 Low ? ? Low Low Low Low
Anna Mou 2021 Low ? Low Low Low Low Low
Yong Tae Kim 2022 ? ? Low Low ? Low Low
Ryohei Yukimoto 2022 Low ? Low Low Low Low Low
Shin Emoto 2023 Low ? Low Low Low Low Low
Rebecca Engel 2024 Low Low Low Low Low Low Low
Tetsuro Kawazoe 2025 Low ? Low Low Low Low Low
Jakob Leonhardi 2025 ? Low Low Low Low Low Low
Open in a new tab

Note: ? = Unclear risk; reporting did not allow assessment of potential bias.

The heterogeneity across studies was substantial (I2 = 73.4%, 95% CI: 61.6%–81.5%; τ2 = 0.327; τ = 0.5719; H = 1.94), indicating a considerable between‐study variability in diagnostic performance. Deeks' test indicated no small‐study effects (bias estimate = 0.755; p = 0.8776).

3.3. Diagnostic Accuracy in All Eligible Cases

A total of 5634 patients were included in the overall analysis. The diagnostic outcomes of all included studies, including sensitivity, specificity, and corresponding 2 × 2 contingency table data, are summarized in (Table 3). Across all eligible studies, the pooled sensitivity of preoperative CT for diagnosing LNM was 0.693 (95% CI: 0.636–0.744), and the pooled specificity was 0.660 (95% CI: 0.581–0.731) (Figure 2).

TABLE 3.

Summary of study characteristics and diagnostic outcomes for all included reports.

No Author Cut‐off value Sensitivity Specificity PPV NPV TP FP FN TN
1 Emil J. Balthazar Max diameter ≥ 10 mm or clusters of ≥ 3 73 58 79 50 44 12 17 17
2 Hiromi Okizuka Max diameter > 8 mm 82 92 90 85 10 1 2 15
3 W Hundt Max diameter ≥ 10 mm or clusters of ≥ 3 and 100 HU 84.3 60 93.1 37.5 27 2 5 3
4 T. Akiyoshi Max diameter > 5 mm 89 52 76 73 25 8 3 20
5 Shuichiro Uchiyama Max diameter > 5 mm 50 69.2 46.7 72 7 8 7 18
6 Mustafa Duman Max diameter ≥ 5 mm or clusters of ≥ 3 59.5 61.1 61.1 59.5 22 14 15 22
7 I‐ H. lao Max diameter ≥ 10 mm or clusters of ≥ 3 70.2 79.2 85.7 60 66 11 30 45
8 M. Lim Max diameter ≥ 10 mm or enhancement 75 62 64 73 49 27 16 44
9 F.E.E. de Vries Max diameter > 10 mm or clusters of ≥ 3 71 41 47 66 32 36 13 25
10 Anne Nørgaard Max diameter > 5 mm 81 26 51 59 29 28 7 10
11 Funda Dinç Elibol max short diameter ≥ 5 mm 84 56 65 78 81 43 15 55
12 Audrey H Choi Max diameter > 10 mm or 7–10 mm + morphology 88 58 59 88 23 16 3 22
13 Armin Wiegering Max short diameter ≥ 5 mm 77.6 46.6 60.1 66.7 83 55 24 48
14 Yan‐qi Huang Max diameter > 10 mm or clusters of ≥ 3 57.4 69.7 65.9 61.6 58 30 43 69
15 Rollvén Erik Max short diameter ≥ 5 mm 90 31 39 86 35 55 4 25
16 Rong‐jie Bai Max diameter > 10 mm 80.8 88 77.8 89.8 21 6 5 44
17 Kazuhito Sasaki Max short diameter > 10 mm 38.0 95.6 90.2 59.2 73 8 118 171
18 Joo Young Lee Max diameter ≥ 10 mm 58.7 64.8 54.5 68.6 59 44 40 78
19 Tsutomu Kumamoto Max long diameter ≥ 7.05 mm 82.1 62.4 55.1 88 23 18 5 24
20 Li‐Heng Liu Max diameter ≥ 7 mm or morphology 72.6 57.4 49.2 79.2 31 32 11 42
21 Shi‐Song Miao Max long diameter ≥ 10 mm 47.0 80.9 48.1 80.2 47 51 53 133
22 Aydin Eresen Max diameter > 10 mm 83.9 47.1 61 74.7 130 83 25 74
23 Anna Mou Max long diameter > 10 mm or < 10 mm + morphology 71.2 74.1 72.5 72.7 37 14 15 40
24 Yong Tae Kim Max short diameter ≥ 5 mm 72.0 52.3 46.7 76.4 134 153 52 168
25 Ryohei Yukimoto Max short diameter ≥ 7 mm 75.2 82.6 42.6 95.1 70 23 94 354
26 Shin Emoto Max short diameter ≥ 5 mm 63.2 53.6 48.0 68.2 100 108 58 125
27 Rebecca Engel Max diameter ≥ 10 mm or 7‐9 mm + morphology 47.92 75.94 58.97 66.89 56 35 49 89
28 Tetsuro Kawazoe Max short diameter ≥ 5.5 mm 75 72.5 62 82.9 185 118 64 327
29 Jakob Leonhardi Max short diameter ≥ 10 mm 39.5 87.7 68 68.7 17 8 26 57
Open in a new tab

Abbreviations: FN, false negative; FP, false positive; NPV, negative predictive value; PPV, positive predictive value; TN, true negative; TP, true positive.

FIGURE 2.

FIGURE 2

Open in a new tab

Forest plots of sensitivity and specificity for preoperative CT in diagnosing lymph node metastasis.

To visually summarize the diagnostic performance, a HS‐ROC curve was constructed using a bivariate random‐effects model (Figure 3). The HS‐ROC curve demonstrated an area under the curve (AUC) of 0.727, indicating a moderate diagnostic accuracy. The summary estimate was accompanied by 95% confidence and prediction regions, reflecting between‐study heterogeneity.

FIGURE 3.

FIGURE 3

Open in a new tab

HS‐ROC curve of preoperative CT for assessing lymph node metastasis.

The DOR was 4.607 (95% CI: 3.545–5.987), while the positive and negative likelihood ratios were 2.018 (95% CI: 1.746–2.332) and 0.505 (95% CI: 0.449–0.568), respectively, further supporting the modest discriminative ability of CT in identifying LNM.

3.4. Comparison of Diagnostic Accuracy Between Studies Focusing Solely on Colon Cancer and Those Including Rectal Cancer

To visualize potential differences in diagnostic performance, we created HS‐ROC curves for studies evaluating only CC (Figure 4a) and for studies analyzing CRC including rectal lesions (excluding studies focused solely on CC) (Figure 4b). In this review, ten studies assessed colon and rectal cancers together, whereas nineteen studies focused exclusively on CC.

FIGURE 4.

FIGURE 4

Open in a new tab

HS‐ROC curves comparing diagnostic accuracy between studies focusing exclusively on CC (a) and independent studies evaluating CRC, excluding CC‐only studies (b).

The shapes and positions of the HS‐ROC curves demonstrated substantial overlap between the two groups, indicating similar diagnostic behavior. Consistently, bivariate meta‐regression showed no significant difference in sensitivity between CRC and CC (p = 0.561). The relative sensitivity ratio (CRC vs. CC) could not be precisely estimated because the regression coefficient was close to zero; however, the 95% confidence interval (0.49–1.48) encompassed unity and suggested no meaningful difference in sensitivity.

Likewise, no significant difference in specificity was detected (p = 0.316), with a wide but uninformative confidence interval (0.34–1.42). Taken together with the visual similarity of the HS‐ROC curves, these findings indicate that the inclusion of rectal cancer did not materially alter the diagnostic accuracy of CT for LNM.

3.5. Comparison of Diagnostic Accuracy Between Studies Focusing Solely on Lymph Node Size and Those Incorporating Morphological Evaluation

To examine whether incorporating morphological evaluation improved diagnostic performance, HS‐ROC curves were generated separately for studies using size criteria alone (Figure 5a) and for studies incorporating morphologic evaluation in addition to size, excluding size‐only studies (Figure 5b). In this review, 18 studies evaluated lymph node size as the sole criterion, while 11 studies included additional morphological evaluation.

FIGURE 5.

FIGURE 5

Open in a new tab

HS‐ROC curves comparing diagnostic accuracy between studies using size‐based CT criteria alone (a) and studies incorporating morphologic assessment in addition to size, excluding size‐only studies (b).

Visual inspection of the HS‐ROC curves revealed considerable overlap between the two groups, suggesting comparable diagnostic behavior. Consistently, bivariate meta‐regression demonstrated no significant difference in sensitivity between size‐based and morphologic CT criteria (p = 0.822), with the relative sensitivity ratio showing a wide but non‐significant range (95% CI: 0.62–1.82).

Similarly, no significant difference in specificity was observed (p = 0.536), and the corresponding relative specificity ratio also crossed unity (95% CI: 0.62–2.54). These findings indicate that the addition of morphological evaluation to size‐based CT criteria did not materially enhance diagnostic accuracy for detecting LNM.

3.6. Sensitivity Analyses

To evaluate the robustness of our findings, we conducted multiple sensitivity analyses. First, the analysis was restricted to studies using ≥ 64‐slice MDCT. The pooled sensitivity and specificity were 0.671 (95% CI, 0.589–0.744) and 0.665 (95% CI, 0.545–0.767), respectively, which were comparable to the results of the main analysis (0.693 [95% CI, 0.636–0.744] and 0.660 [95% CI, 0.581–0.731], respectively), indicating that the diagnostic performance was not materially influenced by CT detector configuration.

Second, we included only studies in which image interpretation was performed by radiologists. This restriction yielded a modestly higher pooled sensitivity of 0.710 (95% CI, 0.651–0.762) but a lower pooled specificity of 0.605 (95% CI, 0.539–0.668) compared with the main analysis, suggesting that radiologist interpretation may enhance the detection of metastatic nodes while increasing the likelihood of false‐positive assessments. In exploratory descriptive analyses stratified by measurement axis and size cutoff, sensitivity and specificity showed substantial variability across thresholds. No single cutoff value consistently demonstrated a superior balance between sensitivity and specificity (Table S1).

4. Discussion

Preoperative CT is widely used as the primary imaging modality for CC and provides essential anatomical information for surgical planning and clinical decision‐making. This systematic review and meta‐analysis demonstrated that CT has only moderate diagnostic accuracy for detecting LNM, with pooled sensitivity and specificity of 0.693 and 0.660, respectively. The HS‐ROC analysis yielded an AUC of 0.727, and the corresponding likelihood ratios and diagnostic odds ratio similarly indicated limited discriminative capability.

Importantly, technological advances have not led to substantial improvements in lymph node assessment. A previous systematic review by Nerad et al. in 201610 reported pooled sensitivity of 71% (95% CI, 59%–81%) and specificity of 67% (95% CI, 46%–83%), with a diagnostic odds ratio of 4.8 (95% CI, 2.5–9.4). Despite including studies published only up to 2014, these estimates are comparable to, or slightly higher than those observed in our analysis. Thus, even with the widespread adoption of 64‐slice or higher MDCT and notable advances such as CT colonography for tumor depth assessment [45, 46], accurate preoperative detection of LNM remains challenging.

Subgroup analyses confirmed that diagnostic limitations were not attributable to specific imaging conditions. Exclusion of studies involving rectal cancer, studies incorporating morphological evaluation, or restriction to 64‐slice or higher MDCT did not improve diagnostic performance. Restricting image interpretation to radiologists slightly improved sensitivity, but the difference was small. Size criteria were not uniformly defined across studies, indicating that achieving accurate diagnosis based solely on size is inherently difficult. Taken together, these findings suggest that despite advances in CT hardware and image reconstruction techniques, the diagnostic performance of CT for detecting LNM in CC appears to have largely plateaued.

Descriptive comparison across size thresholds demonstrated a clear trade‐off between sensitivity and specificity rather than a consistently superior cutoff value. Within the same measurement axis, ranges of sensitivity and specificity overlapped across thresholds; higher cutoffs (e.g., short‐axis ≥ 10 mm) were associated with higher specificity but markedly reduced sensitivity, whereas lower cutoffs (e.g., ≥ 5 mm) increased sensitivity at the expense of specificity. Consequently, the available data do not identify a single size threshold that provides a clearly optimal balance between sensitivity and specificity.

Therefore, alternative imaging strategies to improve accuracy are needed. Node‐RADS [47] was developed to integrate texture, margin, and shape in addition to size, but its current clinical application is limited due to significant interobserver variability reported in colorectal cancer [44]. PET/CT also holds potential for improving diagnostic accuracy. For example, Yukimoto et al. [40] reported a sensitivity of 78.6% and specificity of 96.8% when using a lymph node SUVmax cutoff value of 1.5. However, practical issues such as radiation exposure and cost limit its widespread use in routine staging.

These diagnostic challenges significantly influence treatment decisions. Even if colectomy itself is routinely performed, inaccurate preoperative nodal staging may influence the planned extent of lymph node dissection, perioperative treatment strategies, patient selection for neoadjuvant therapy, and interpretation of oncological risk. In Japan, when regional LNM is clinically suspected, Japanese guidelines [5] recommend D3 lymph node dissection, with preoperative CT findings playing a major role in determining surgical strategy. Furthermore, lymph node staging may influence decisions regarding preoperative chemotherapy. A report by Mizuuchi et al. [48] demonstrated that N3 LNM is an independent poor prognostic factor in patients with Stage III CC. While the FOxTROT [49] and OPTICAL [7] trials used only the T factor for preoperative chemotherapy eligibility, the JCOG2006 [50] trial included cT3–T4a and positive LNM (cN2 or cN3) in its eligibility criteria. Whether the cN factor is useful as an indicator for initiating preoperative chemotherapy requires further investigation.

Taken together, these findings highlight the need for an international standardized approach to LNM assessment. Limited studies have evaluated MRI for assessing LNM in CC. According to Bompou et al. [51], MRI showed a sensitivity of 100% for two interpreters, with specificities of 90.2% and 95.2%, respectively, and significantly higher AUC values than CT (CT: 0.763 and 0.751; MRI: 0.951 and 0.976; p < 0.001). On the other hand, Rafaelsen et al. [52] reported MRI sensitivity of 58.2% and specificity of 50.0%, showing no significant difference from CT. Therefore, further investigation is needed regarding the utility of MRI. Technologies such as radiomics and artificial intelligence show potential to overcome the constraints of size‐based diagnosis. In a radiomics study by Eresen [37] et al., a model derived from radiological image data demonstrated an accuracy of 81.09% and 79.49% in the training and test cohorts, respectively, with sensitivities of 83.87% and 74.36%, and specificities of 78.34% and 84.62%. In the rapidly advancing field of AI, automated CT‐based models achieved an AUC of 0.92 for predicting LNM in pancreatic ductal adenocarcinoma, significantly surpassing the AUC of 0.65 achieved by radiologists [53]. Similarly, in bladder cancer, AI systems trained on whole‐slide pathology images demonstrated significantly higher diagnostic sensitivity than junior pathologists (0.983 vs. 0.906) and senior pathologists (0.986 vs. 0.947), while also reducing the rate of missed micrometastases [54]. These advances suggest that AI‐assisted image interpretation may pave the way for improved and standardized lymph node assessment in the future.

This review has several limitations. The included studies differed in lymph node measurement techniques, including the anatomical site of assessment, the choice of short‐ versus long‐axis diameter, and the cutoff values used to define metastatic involvement. Specifically, 9 studies used the short axis as the maximum diameter, while 20 studies used the long axis. The most common maximum diameter was 10 mm, used in 15 studies, followed by 5 mm in 9 studies and 7 mm in 3 studies. The number of such studies further decreased that measured size alone without morphological evaluation, making it difficult to compare which size yielded the highest AUC. This variability prevented standardized comparison of diagnostic performance across specific size thresholds. Furthermore, studies evaluating only morphological features without reporting size were excluded, limiting the quantitative comparison between size‐based CT criteria and morphology‐based CT criteria. However, recent studies that do not include size measurements are limited. Since the recent studies by Rollvén et al. [30] targeted Stage III and Li et al. [55] targeted T1, there may be no substantial difference in results.

In conclusion, CT using size criteria alone for preoperative lymph node staging in colon cancer demonstrated only moderate accuracy even in the era of 64‐slice or higher MDCT. These findings indicate that relying on lymph node size as the primary determinant remains insufficient for reliable clinical decision‐making. Standardized evaluation methods and advanced imaging strategies may be necessary to improve diagnostic performance.

Author Contributions

Yuji Takayama: conceptualization, methodology, data curation, formal analysis, writing – original draft, writing – review and editing, visualization. Tomoya Shimizu: methodology, data curation, validation. Jun Watanabe: methodology, conceptualization, formal analysis, supervision. Shimpei Ogawa: methodology, validation, supervision. Yukihide Kanemitsu: project administration, supervision.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Descriptive summary of diagnostic performance stratified by measurement axis and size cutoff.

AGS3-10-1450-s002.docx (17.1KB, docx)

Data S1: Supplementary: Literature Search Strategy.

AGS3-10-1450-s001.docx (16.3KB, docx)

Acknowledgments

The authors thank the members of the Lymph Node Committee of the Japanese Society for Cancer of the Colon and Rectum for their valuable contributions.

References

  • 1. Chapuis P. H., Dent O. F., Fisher R., et al., “A Multivariate Analysis of Clinical and Pathological Variables in Prognosis After Resection of Large Bowel Cancer,” Journal of British Surgery 72, no. 9 (1985): 698–702. [DOI] [PubMed] [Google Scholar]
  • 2. Wolmark N., Fisher B., and Wieand H. S., “The Prognostic Value of the Modifications of the Dukes' C Class of Colorectal Cancer. An Analysis of the NSABP Clinical Trials,” Annals of Surgery 203, no. 2 (1986): 115–122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Cohen A. M., Tremiterra S., Candela F., Thaler H. T., and Sigurdson E. R., “Prognosis of Node‐Positive Colon Cancer,” Cancer 67, no. 7 (1991): 1859–1861. [DOI] [PubMed] [Google Scholar]
  • 4. Suzuki O., Sekishita Y., Shiono T., Ono K., Fujimori M., and Kondo S., “Number of Lymph Node Metastases Is Better Predictor of Prognosis Than Level of Lymph Node Metastasis in Patients With Node‐Positive Colon Cancer,” Journal of the American College of Surgeons 202, no. 5 (2006): 732–736. [DOI] [PubMed] [Google Scholar]
  • 5. Kinugasa Y., Uehara K., Yamaguchi K., et al., “Japanese Society for Cancer of the Colon and Rectum (JSCCR) Guidelines 2024 for the Treatment of Colorectal Cancer,” International Journal of Clinical Oncology 30 (2025): 2410–2463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Foxtrot Collaborative Group , “Feasibility of Preoperative Chemotherapy for Locally Advanced, Operable Colon Cancer: The Pilot Phase of a Randomised Controlled Trial,” Lancet Oncology 13, no. 11 (2012): 1152–1160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Hu H., Zhang J., Li Y., et al., “Neoadjuvant Chemotherapy With Oxaliplatin and Fluoropyrimidine Versus Upfront Surgery for Locally Advanced Colon Cancer: The Randomized, Phase III OPTICAL Trial,” Journal of Clinical Oncology 42, no. 25 (2024): 2978–2988. [DOI] [PubMed] [Google Scholar]
  • 8. Hofheinz R. D., Fokas E., Benhaim L., et al., “Localised Rectal Cancer: ESMO Clinical Practice Guideline for Diagnosis, Treatment and Follow‐Up,” Annals of Oncology 36, no. 9 (2025): 1007–1024. [DOI] [PubMed] [Google Scholar]
  • 9. Benson A. B., Venook A. P., Adam M., et al., “NCCN Guidelines Insights: Rectal Cancer, Version 3.2024,” Journal of the National Comprehensive Cancer Network 22, no. 6 (2024): 366–375. [DOI] [PubMed] [Google Scholar]
  • 10. Nerad E., Lahaye M. J., Maas M., et al., “Diagnostic Accuracy of CT for Local Staging of Colon Cancer: A Systematic Review and Meta‐Analysis,” AJR. American Journal of Roentgenology 207, no. 5 (2016): 984–995. [DOI] [PubMed] [Google Scholar]
  • 11. McInnes M. D. F., Moher D., Thombs B. D., McGrath T. A., and Bossuyt P. M., “Preferred Reporting Items for a Systematic Review and Meta‐Analysis of Diagnostic Test Accuracy Studies: The PRISMA‐DTA Statement,” Journal of the American Medical Association 319, no. 4 (2018): 388–396. [DOI] [PubMed] [Google Scholar]
  • 12. Whiting P. F., Rutjes A. W. S., Westwood M. E., et al., “QUADAS‐2: A Revised Tool for the Quality Assessment of Diagnostic Accuracy Studies,” Annals of Internal Medicine 155, no. 8 (2011): 529–536. [DOI] [PubMed] [Google Scholar]
  • 13. Rutter C. M. and Gatsonis C. A., “A Hierarchical Regression Approach to Meta‐Analysis of Diagnostic Test Accuracy Evaluations,” Statistics in Medicine 20, no. 19 (2001): 2865–2884. [DOI] [PubMed] [Google Scholar]
  • 14. Reitsma J. B., Glas A. S., Rutjes A. W. S., Scholten R. J. P. M., Bossuyt P. M., and Zwinderman A. H., “Bivariate Analysis of Sensitivity and Specificity Produces Informative Summary Measures in Diagnostic Reviews,” Journal of Clinical Epidemiology 58, no. 10 (2005): 982–990. [DOI] [PubMed] [Google Scholar]
  • 15. Deeks J. J., Macaskill P., and Irwig L., “The Performance of Tests of Publication Bias and Other Sample Size Effects in Systematic Reviews of Diagnostic Test Accuracy Was Assessed,” Journal of Clinical Epidemiology 58, no. 9 (2005): 882–893. [DOI] [PubMed] [Google Scholar]
  • 16. Balthazar E. J., Megibow A. J., Hulnick D., and Naidich D. P., “Carcinoma of the Colon: Detection and Preoperative Staging by CT,” AJR. American Journal of Roentgenology 150, no. 2 (1988): 301–306. [DOI] [PubMed] [Google Scholar]
  • 17. Okizuka H., Sugimura K., Shinozaki N., and Watanabe K., “Colorectal Carcinoma: Evaluation With Ultrafast CT,” Clinical Imaging 19, no. 4 (1995): 247–251. [DOI] [PubMed] [Google Scholar]
  • 18. Hundt W., Braunschweig R., and Reiser M., “Evaluation of Spiral CT in Staging of Colon and Rectum Carcinoma,” European Radiology 9, no. 1 (1999): 78–84. [DOI] [PubMed] [Google Scholar]
  • 19. Akiyoshi T., Oya M., Fujimoto Y., et al., “Comparison of Preoperative Whole‐Body Positron Emission Tomography With MDCT in Patients With Primary Colorectal Cancer,” Colorectal Disease 11, no. 5 (2009): 464–469. [DOI] [PubMed] [Google Scholar]
  • 20. Uchiyama S., Haruyama Y., Asada T., Hotokezaka M., Nagamachi S., and Chijiiwa K., “Role of the Standardized Uptake Value of 18‐Fluorodeoxyglucose Positron Emission Tomography‐Computed Tomography in Detecting the Primary Tumor and Lymph Node Metastasis in Colorectal Cancers,” Surgery Today 42, no. 10 (2012): 956–961. [DOI] [PubMed] [Google Scholar]
  • 21. Duman M., Tas S., Mecit E. A., et al., “Preoperative Local Staging of Colorectal Cancer Patients With MDCT,” Hepato‐Gastroenterology 59, no. 116 (2012): 1108–1112. [DOI] [PubMed] [Google Scholar]
  • 22. Lao I. H., Chao H., Wang Y. J., et al., “Computed Tomography Has Low Sensitivity for the Diagnosis of Early Colon Cancer,” Colorectal Disease 15, no. 7 (2013): 807–811. [DOI] [PubMed] [Google Scholar]
  • 23. Lim M., Hussain Z., Howe A., et al., “The Oncological Outcome After Right Hemicolectomy and Accuracy of CT Scan as a Preoperative Tool for Staging in Right Sided Colonic Cancers,” Colorectal Disease 15, no. 5 (2013): 536–543. [DOI] [PubMed] [Google Scholar]
  • 24. de Vries F. E. E., da Costa D. W., van der Mooren K., van Dorp T. A., and Vrouenraets B. C., “The Value of Pre‐Operative Computed Tomography Scanning for the Assessment of Lymph Node Status in Patients With Colon Cancer,” European Journal of Surgical Oncology 40, no. 12 (2014): 1777–1781. [DOI] [PubMed] [Google Scholar]
  • 25. Nørgaard A., Dam C., Jakobsen A., Pløen J., Lindebjerg J., and Rafaelsen S. R., “Selection of Colon Cancer Patients for Neoadjuvant Chemotherapy by Preoperative CT Scan,” Scandinavian Journal of Gastroenterology 49, no. 2 (2014): 202–208. [DOI] [PubMed] [Google Scholar]
  • 26. Elibol F. D., Obuz F., Sökmen S., et al., “The Role of Multidetector CT in Local Staging and Evaluation of Retroperitoneal Surgical Margin Involvement in Colon Cancer,” Diagnostic and Interventional Radiology 22, no. 1 (2016): 5–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Choi A. H., “Accuracy of Computed Tomography in Nodal Staging of Colon Cancer Patients,” World Journal of Gastrointestinal Surgery 7, no. 7 (2015): 116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Wiegering A., Kunz M., Hussein M., et al., “Diagnostic Value of Preoperative CT Scan to Stratify Colon Cancer for Neoadjuvant Therapy,” International Journal of Colorectal Disease 30, no. 8 (2015): 1067–1073. [DOI] [PubMed] [Google Scholar]
  • 29. Huang Y. Q., Liang C. H., He L., et al., “Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer,” Journal of Clinical Oncology 34, no. 18 (2016): 2157–2164. [DOI] [PubMed] [Google Scholar]
  • 30. Rollvén E., Abraham‐Nordling M., Holm T., and Blomqvist L., “Assessment and Diagnostic Accuracy of Lymph Node Status to Predict Stage III Colon Cancer Using Computed Tomography,” Cancer Imaging 17, no. 1 (2017): 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Bai R. J., Ren S. H., Jiang H. J., Li J. P., Liu X. C., and Xue L. M., “Accuracy of Multi‐Slice Spiral Computed Tomography for Preoperative Tumor Node Metastasis (TNM) Staging of Colorectal Carcinoma,” Medical Science Monitor 23 (2017): 3470–3479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Sasaki K., Kawasaki H., Sato M., Koyama K., Yoshimi F., and Nagai H., “Impact of 18F‐Fluorodeoxyglucose Uptake on Preoperative Positron Emission Tomography/Computed Tomography in the Lymph Nodes of Patients With Primary Colorectal Cancer,” Digestive Surgery 34, no. 1 (2017): 60–67. [DOI] [PubMed] [Google Scholar]
  • 33. Lee J. Y., Yoon S. M., Kim J. T., et al., “Diagnostic and Prognostic Value of Preoperative 18‐Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography for Colorectal Cancer: Comparison With Conventional Computed Tomography,” Intestinal Research 15, no. 2 (2017): 208–214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Kumamoto T., Shindoh J., Mita H., et al., “Optimal Diagnostic Method Using Multidetector‐Row Computed Tomography for Predicting Lymph Node Metastasis in Colorectal Cancer,” World Journal of Surgical Oncology 17, no. 1 (2019): 39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35. Liu L. H., Lv H., Wang Z. C., Rao S. X., and Zeng M. S., “Performance Comparison Between MRI and CT for Local Staging of Sigmoid and Descending Colon Cancer,” European Journal of Radiology 121 (2019): 108741. [DOI] [PubMed] [Google Scholar]
  • 36. Miao S. S., Lu Y. F., Chen H. Y., et al., “Contrast‐Enhanced CT Imaging for the Assessment of Lymph Node Status in Patients With Colorectal Cancer,” Oncology Letters 19, no. 5 (2020): 3451–3458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Eresen A., Li Y., Yang J., et al., “Preoperative Assessment of Lymph Node Metastasis in Colon Cancer Patients Using Machine Learning: A Pilot Study,” Cancer Imaging 20, no. 1 (2020): 30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Mou A., Li H., Chen X. L., Fan Y. H., and Pu H., “Tumor Size Measured by Multidetector CT in Resectable Colon Cancer: Correlation With Regional Lymph Node Metastasis and N Stage,” World Journal of Surgical Oncology 19 (2021): 302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Kim Y. T., Min J. H., Choi K. H., and Kim H., “Colon Cancer Microsatellite Instability Influences Computed Tomography Assessment of Regional Lymph Node Morphology and Diagnostic Performance,” European Journal of Radiology 154 (2022): 110396. [DOI] [PubMed] [Google Scholar]
  • 40. Yukimoto R., Uemura M., Tsuboyama T., et al., “Efficacy of PET/CT in Diagnosis of Regional Lymph Node Metastases in Patients With Colorectal Cancer: Retrospective Cohort Study,” BJS Open 6, no. 4 (2022): zrac090, 10.1093/bjsopen/zrac090. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Emoto S., Homma S., Ichikawa N., et al., “Preoperative Prediction of Lymph Node Metastasis From Colorectal Cancer Using Computed Tomography,” SN Comprehensive Clinical Medicine 5 (2023): 115. [Google Scholar]
  • 42. Engel R., Kudura K., Antwi K., et al., “Diagnostic Accuracy and Treatment Benefit of PET/CT in Staging of Colorectal Cancer Compared to Conventional Imaging,” Surgical Oncology 57 (2024): 102151. [DOI] [PubMed] [Google Scholar]
  • 43. Kawazoe T., Nakanishi R., Ando K., Zaitsu Y., Kudou K., and Nakashima Y., “Preoperative CT Lymph Node Size as a Predictor of Nodal Metastasis in Resectable Colon Cancer: A Retrospective Study of 694 Patients,” BMC Gastroenterology 25, no. 1 (2025): 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. Leonhardi J., Mehdorn M., Stelzner S., et al., “Diagnostic Accuracy and Reliability of CT‐Based Node‐RADS for Colon Cancer,” Abdom Radiol N Y 50, no. 1 (2025): 1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Hoshino N., Sakamoto T., Hida K., and Sakai Y., “Diagnostic Accuracy of Computed Tomography Colonography for Tumor Depth in Colorectal Cancer: A Systematic Review and Meta‐Analysis,” Surgical Oncology 30 (2019): 126–130. [DOI] [PubMed] [Google Scholar]
  • 46. Komono A., Shida D., Iinuma G., Tsukamoto S., Sakamoto R., and Moritani K., “Preoperative T Staging of Colon Cancer Using CT Colonography With Multiplanar Reconstruction: New Diagnostic Criteria Based Bordering Vessels,” International Journal of Colorectal Disease 34, no. 4 (2019): 641–648. [DOI] [PubMed] [Google Scholar]
  • 47. Elsholtz F. H. J., Asbach P., Haas M., et al., “Introducing the Node Reporting and Data System 1.0 (Node‐RADS): A Concept for Standardized Assessment of Lymph Nodes in Cancer,” European Radiology 31, no. 8 (2021): 6116–6124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Mizuuchi Y., Tanabe Y., Sada M., et al., “Relationship Between Prognostic Impact of N3 Lymph Node Metastasis at the Root of the Feeding Artery and Location of Colon Cancer,” Langenbeck's Archives of Surgery 408, no. 1 (2023): 31. [DOI] [PubMed] [Google Scholar]
  • 49. Morton D., Seymour M., Magill L., et al., “Preoperative Chemotherapy for Operable Colon Cancer: Mature Results of an International Randomized Controlled Trial,” Journal of Clinical Oncology 41, no. 8 (2023): 1541–1552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Hashimoto T., Maruyama S., Takii Y., et al., “A Randomized Phase II Study Comparing Preoperative mFOLFOX6 Versus FOLFOXIRI for Locally Advanced Colon Cancer: JCOG2006,” Future Oncology 19, no. 28 (2023): 1897–1904. [DOI] [PubMed] [Google Scholar]
  • 51. Bompou E., Vassiou A., Baloyiannis I., Perivoliotis K., Fezoulidis I., and Tzovaras G., “Comparative Evaluation of CT and MRI in the Preoperative Staging of Colon Cancer,” Scientific Reports 14, no. 1 (2024): 17145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Rafaelsen S. R., Dam C., Vagn‐Hansen C., et al., “CT and 3 Tesla MRI in the TN Staging of Colon Cancer: A Prospective, Blind Study,” Current Oncology 29, no. 2 (2022): 1069–1079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53. Bian Y., Zheng Z., Fang X., et al., “Artificial Intelligence to Predict Lymph Node Metastasis at CT in Pancreatic Ductal Adenocarcinoma,” Radiology 306, no. 1 (2023): 160–169. [DOI] [PubMed] [Google Scholar]
  • 54. Wu S., Hong G., Xu A., et al., “Artificial Intelligence‐Based Model for Lymph Node Metastases Detection on Whole Slide Images in Bladder Cancer: A Retrospective, Multicentre, Diagnostic Study,” Lancet Oncology 24, no. 4 (2023): 360–370. [DOI] [PubMed] [Google Scholar]
  • 55. Li S., Li Z., Wang L., et al., “CT Morphological Features for Predicting the Risk of Lymph Node Metastasis in T1 Colorectal Cancer,” European Radiology 33, no. 10 (2023): 6861–6871. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1: Descriptive summary of diagnostic performance stratified by measurement axis and size cutoff.

AGS3-10-1450-s002.docx (17.1KB, docx)

Data S1: Supplementary: Literature Search Strategy.

AGS3-10-1450-s001.docx (16.3KB, docx)

Articles from Annals of Gastroenterological Surgery are provided here courtesy of Wiley

RESOURCES