Clinical TNM staging for esophageal, gastric, and colorectal cancers in the era of neoadjuvant therapy: A systematic review of the literature

Abstract Aim Clinical staging is vital for selecting appropriate candidates and designing neoadjuvant treatment strategies for advanced tumors. The aim of this review was to evaluate diagnostic abilities of clinical TNM staging for gastrointestinal, gastrointestinal cancers. Methods We conducted a systematic review of recent publications to evaluate the accuracy of diagnostic modalities on gastrointestinal cancers. A systematic literature search was performed in PubMed/MEDLINE using the keywords “TNM staging,” “T4 staging,” “distant metastases,” “esophageal cancer,” “gastric cancer,” and “colorectal cancer,” and the search terms used in Cochrane Reviews between January 2005 to July 2020. Articles focusing on preoperative diagnosis of: (a) depth of invasion; (b) lymph node metastases; and (c) distant metastases were selected. Results After a full‐text search, a final set of 55 studies (17 esophageal cancer studies, 26 gastric cancer studies, and 12 colorectal cancer studies) were used to evaluate the accuracy of clinical TNM staging. Positron emission tomography–computed tomography (PET‐CT) and/or magnetic resonance imaging (MRI) were the best modalities to assess distant metastases. Fat and fiber mode of CT may be useful for T4 staging of esophageal cancer, CT was a partially reliable modality for lymph node staging in gastric cancer, and CT combined with MRI was the most reliable modality for liver metastases from colorectal cancer. Conclusion The most reliable diagnostic modality differed among gastrointestinal cancers depending on the type of cancer. Therefore, we propose diagnostic algorithms for clinical staging for each type of cancer.


| INTRODUC TI ON
Despite recent advances in surgical techniques, perioperative care, and multimodal treatment, postoperative recurrence is observed in approximately 40% of esophageal cancers 1 and 20%-30% of gastric 2 and colorectal 3 cancers with advanced tumors. [4][5][6] Lymph node metastasis remains crucial for applying adjuvant treatment and predicting oncological outcome. Various studies have shown that such postoperative recurrence was frequently reduced by neoadjuvant chemotherapy (NAC). [7][8][9][10][11] The present review evaluates the accuracy of preoperative diagnosis in gastrointestinal cancers, including lymph node metastasis.
Potential T4 esophageal cancer should be treated with neoadjuvant chemoradiation therapy to ensure a negative surgical margin for cancer cells. NAC became the standard management for stage II/ III esophageal cancer following the results of the JCOG9907 trial. 12 JCOG1002 evaluated NAC for locally advanced gastric cancer with extended lymph node metastasis and/or bulky positive nodes. 13 Two more ongoing trials also evaluating NAC for locally advanced gastric cancer. 14,15 Distant metastases should be classified as a noncurative factor for surgical approach. Therefore, clinical TNM staging should be accurate, based on high sensitivity and specificity to predict T4 and/or distant metastases. Since definitive chemoradiation therapy showed a similar overall survival to radical surgery for clinical stage I esophageal cancer, 16 an accurate diagnosis of lymph node metastases is also vital to design treatment strategies for potential stage I tumors.
In Western countries, preoperative chemotherapy or chemoradiotherapy is a standard therapeutic strategy for advanced gastric cancer, based on the findings from large-scale randomized clinical trials. [17][18][19][20] While advanced stage gastric cancer is the main target of NAC, 8.3% of pathological T1 patients were included in the surgery alone group, 17 indicating that some early gastric cancer patients underwent unnecessary NAC. This problem may be due to inaccuracy of clinical diagnosis of T and N staging. In Japan, the efficacy of NAC for type 4 and large-sized type 3 was not demonstrated in the JCOG0501 trial. 21 The JCOG1302A trial, which evaluated the accuracy of clinical diagnosis of gastric cancer, was conducted as prospective setting prior to starting the JCOG1509 trial 22 regarding the efficacy of NAC for stage III gastric cancer. 23 The JCOG1310 trial (PRECIOUS study) is intended to compare preoperative vs postoperative chemotherapy for lower rectal cancer patients with suspected lateral pelvic node metastasis. 24 MRI has been reported to be the most effective tool for the preoperative stage diagnosis of rectal cancer. 25 It remains controversial whether chemotherapy with or without primary tumor resection is effective for patients with incurable stage IV colorectal cancer. The precise detection of distant metastases 26 is vital in order to enroll patients for such a typical randomized study.
Thus, the impact of clinical TNM staging is more important than ever since neoadjuvant therapy for gastrointestinal cancers is becoming established. Therefore, we evaluated the accuracy of clinical TNM staging through multimodal diagnostic tools using a systematic review of recent publications from January 2005 to July 2020. We propose the use of standard diagnostic algorithms for gastrointestinal cancers. The present review aimed to summarize the fundamental information about the accuracy of clinical TNM staging to design future guidelines and clinical protocols for preoperative adjuvant therapy for gastrointestinal cancers.

| Research themes and study selection criteria
The present review focused on esophageal, gastric, and colorec- July 2020. In PubMed, the search terms "esophageal cancer," "gastric cancer," "colorectal cancer," and "TNM staging" were used. In MEDLINE, the search terms used in Cochrane Reviews were used (advanced search system, Appendix 1). 27 The relevance of each ar-

| Data extraction
Key messages and information were extracted from each article and organized. The following information from eligible articles was used: authors, title, countries of origin, publication year, total sample size, study design, study period, diagnostic modality, conclusion, and summary statistics (sensitivity, specificity, and number of positive and negative patients) for diagnosis. We focused on two statistical measurements of diagnostic accuracy of the modality: sensitivity (the proportion of positively diagnosed patients with disease) and specificity (the proportion of negatively diagnosed patients without disease).

| Studies included in this paper
Our systematic search identified 23  hand, EUS or MRI was appropriate to determine non-T4 status. The most reliable diagnostic modalities include a combination of EUS and CT. Kobayashi et al analyzed the characteristics of the esophageal motion and esophageal internal target volume margins to assess the differences between clinical T1-T3 and clinical T4 using fourdimensional CT. 35 Although the accuracy of EUS was the highest among these diagnostic modalities, CT or MRI were appropriate modalities to detect T4 status. Figure 1 shows the differential diagnosis between T4 and T3 tumors of esophageal cancer after chemoradiation therapy using an image reconstruction method according to the CT value of the tissue histology of enhanced CT, so called fat and fiber mode. 36 In this examination, the contrast agent (3 mL/kg body) was administrated and the legions were scanned with a 50-second delay and a thickness of 1 mm. The fibrotic area induced by the chemoradiation therapy was emphasized as green and the presence of a fibrotic layer between the tumor and the adjacent organs could be interpreted as not T4.

| Lymph node staging
A total of 13 studies in 10 manuscripts 37-46 evaluated the diagnostic impact of EUS/CT/PET on the nodal involvement in esophageal cancer (Table 1B). Four out of 13 eligible studies used combination diagnosis with either PET/CT or EUC/CT. The sensitivity of nodal involvement ranged from 29% to 94%, and the specificity ranged from 38% to 98% (Table 1B).

| Diagnosis for distant metastases
Four studies in three manuscripts 37,47,48 evaluated the diagnostic accuracy of imaging to detect distant metastases (Table 1C). The useful modalities were CT and PET. The advantage of CT was its high resolution to detect the lesion with an accuracy of 77%, 37 while the advantage of PET was the ability to perform whole-body scanning with highly qualitative contrasted metastatic lesions identified by high glucose uptake with an accuracy of 85% ~ 92%. [46][47][48][49][50][51] These accuracies were around 10% greater than that of CT; therefore, CT and PET should be used together as morphological and qualitative modalities.

| Algorithm of image modalities for clinical staging in esophageal cancer
Based on these findings, we suggest an algorithm of image modalities for clinical staging in esophageal cancer ( Figure 2 and 78%, respectively, by Seevaratnam et al, 63 and 67% and 84%, respectively, by Wang et al. 66 The incidence of lymph node metastases was higher in advanced tumors than in early tumors. When limited to T1 tumors, the sensitivities and specificities were reported to be 17% and 90%, respectively, by Ahn et al 58 and 4.3% and 98%, respectively, by Fujikawa et al. 67 When limited to T2-T4, the specificity and sensitivity were reported to be 63% and 66%, respectively, by Fukagawa et al. 21 One of the reasons for difficulties in lymph node diagnosis is the diagnostic difficulty for small-sized lymph node metastases. Figure 3 shows a series of CT images at the same position after endoscopic submucosal dissection (ESD) in a patient who underwent ESD for T1a early gastric cancer. The pathological depth of the resected specimen was sm2, and there was the possibility of simultaneous lymph node metastases. However, this patient chose to be monitored using CT examinations every 6 months without additional surgery for lymph node dissection. The lymph node was found to be clinically metastatic at the No. 6 station. Following this CT finding, the patient underwent distal gastrectomy and one lymph node was found to be pathologically positive for metastasis at the No. 6 station, which was compatible with the CT findings. Looking at these CT images, a tiny lymph node was visible ( Figure 3A,B) in the same area, with swollen node visible ( Figure 3C). This tiny lymph node may have been positive for metastasis at that time but was not found to be clinically positive due to its small size. This patient underwent distal gastrectomy after this CT finding, and one lymph node was pathologically positive for metastasis at No. 6, the same with the CT finding. Looking back at these CT images, a tiny lymph node was visible (in A and B) at the same area with swollen node in (C). This tiny lymph node was positive for metastasis at that time, which was not clinically positive for its small size.

| Diagnosis for distant metastases
Peritoneal dissemination is diagnosed using CT, with findings of ascites and multiple mesenteric or omental nodules; however, its diagnostic accuracy is not high (Table 2C). 68 The standard therapeutic strategy for advanced gastric cancer with peritoneal dissemination is systemic chemotherapy without gastrectomy, as determined by the REGATTA trial. 69 The detection of small peritoneal dissemination by staging laparoscopy can avoid unnecessary laparotomy. A total of 10 manuscripts 70-79 evaluated the diagnostic accuracy of peritoneal metastases by staging laparoscopy (Table 2C). The detection ratio of peritoneal dissemination was found to be 7.8%-36%. 77  The percentage of "false negatives" is reported to be 11%-17% in Japan and 0%-8% in Western countries. 80 The reason for this discrepancy is considered to be the difference in the indication of staging laparoscopy.

Accuracy (%)
A) The summary of diagnostic modalities for T staging in colorectal cancer.   Table 4. The common criteria for metastatic nodes were "round shape" and "enhancement" in gastrointestinal cancers. The optimal cutoff size to classify the positive lymph nodes differed according to the type of cancer as follows: 5-10 mm in esophageal nodes, 96 8-10 mm in gastric nodes, 21 and 4-5 mm in colorectal nodes. 88,97 Although the diagnostic accuracy in esophageal cancer was relatively higher than that in gastric and colorectal cancers, the accuracies in all three types of cancer were unsatisfactory. reports. 103,106 The diagnostic characteristics of submucosal invasion are not described clearly and diagnostic ESD is performed in some cases. In Western countries, the standard therapeutic strategy for advanced gastric cancer is NAC based on the results of pivotal clinical trials, such as the FLOT trial 20 and others. 16 The rate of accurate diagnosis of conventional diagnostic imaging was evaluated in patients who underwent radical surgery without preoperative treatment. However, many advanced cancers will become candidates for preoperative treatment. Therefore, it will be necessary to perform diagnostic imaging before and after preoperative chemotherapy to monitor changes in staging and the rate of agreement with postoperative pathological staging. It is not possible to verify whether pretreatment staging was correct in patients undergoing preoperative chemotherapy. However, if the staging by diagnostic imaging after preoperative treatment matches the postoperative pathological staging, it may be possible to ensure the accuracy of the staging prior to treatment. In the future, more accurate pathological therapeutic effects and staging will be required after preoperative treatment. In patients receiving preoperative treatment, difficulties remain in terms of lymph node metastasis diagnosis and the usefulness of PET is predicted to become more important.

| D ISCUSS I ON
In conclusion, our literature review suggests that the recent diagnostic modalities can make precise differential diagnoses for T4, N1, and M1 for gastrointestinal cancers. However, the accuracy is still not sufficient to design preoperative treatment strategies.
The most important purpose of clinical staging is to determine whether neoadjuvant therapy should be performed on each patient. Overstaging could occur in some patients without a standard algorithm for clinical staging and may lead to overtreatment.
Accurate diagnostic modalities that adhere to a standard algorithm may improve both oncological outcomes and patient quality of life.
Since there are only a few large-scale prospective cohort studies in this field, further multi-institutional prospective studies are required.

ACK N OWLED G EM ENTS
We thank Ms Seiko Otsuka and Yuko Okamura for preparing the data of the selected papers.

D I SCLOS U R E
Funding: This work has been partly supported by a research grant of