Improved prognostic stratification of the revised ninth edition American Joint Committee on Cancer staging system for colorectal cancer

Abstract

BACKGROUND

The 9^th edition American Joint Committee on Cancer (AJCC) staging system has been proposed by the AJCC Cancer Expert Panel (AJCCCCEP). The prognostic discriminatory validity of the new staging system in actual patient cohorts has not been fully validated.

AIM

To investigate and validate the proposed changes to the pathologic staging for colon cancer (CC) and rectal cancer (RC) by the AJCCCCEP.

METHODS

The study included 1251 eligible patients with stage I-III CC and 1162 patients with RC between July 2015 and July 2021. Clinicopathological features of the patients were obtained. Patients were staged according to 8^th edition AJCC staging system and 9^th edition AJCC staging system, respectively. Patients’ outcomes were investigated with respect to overall survival (OS) and disease-free survival (DFS) according to the 8^th edition and 9^th edition, respectively. The Kaplan-Meier curve and log-rank test were conducted to compare outcomes of patients with different stages. Hazard ratio was calculated using the Cox proportional hazard model.

RESULTS

One thousand, two hundred and fifty-one patients with CC and 1162 patients with RC were analyzed, with a median follow-up of 5.45 and 5.30 years for DFS and OS in the CC cohort and 5.36 and 5.35 years in the RC cohort. All patients were re-classified according to the updated staging system proposed by AJCCCCEP. The survival analysis revealed a hierarchical separation across all sub-levels in the CC cohort and the RC cohort in terms of DFS and OS.

CONCLUSION

This study indicates that the 9^th edition AJCC staging system has better prognostic validity than the 8^th edition AJCC staging system for patients with CC and RC.

Key Words: Colon cancer; Rectal cancer; Ninth edition American Joint Committee on Cancer staging system; Prognosis; Colorectal cancer

Core Tip: This validation study of 2413 stage I-III colon and rectal cancer patients demonstrates that the proposed 9^th edition American Joint Committee on Cancer staging system shows clearer hierarchical separation in overall and disease-free survival, offering better prognostic discrimination compared to the 8^th edition.

INTRODUCTION

Colorectal cancer (CRC) is the third most common cancer and the second leading cause of mortality among all cancer types worldwide[1]. The American Joint Committee on Cancer (AJCC) staging manual, which classifies tumor from the perspective of pathological anatomy remains the cornerstone for treatment and prognosis of patients with CRC, for example, stage III patients are routinely given adjuvant chemotherapy, while in stage II patients, a decision is often made on a case-by-case basis. However, the discrimination and accuracy of prognostic competency using AJCC staging criteria remains controversial. The latest 8^th edition AJCC staging system was released in 2016, in which the regional lymph nodes involvement (N component) is considered to be a more important prognostic factor than the depth of tumor invasion into the colorectal wall (T component), whereas numerous previous studies demonstrated that T4N0 colon cancer (CC) had significantly poorer outcomes than T1-2N1-2a CCs, regardless of the number of lymph nodes examined[2-4]. Moreover, the survival curves by the 8^th edition summary stage are not hierarchical, with stage I and stage III patients exhibiting similar overall survival (OS) and stage IIIB patients exhibiting superior OS compared to stage IIB and IIC patients. Recently, the 9^th edition AJCC staging system has been proposed by the AJCC Cancer Expert Panel (AJCCCCEP) and reported in American Society of Clinical Oncology 2025, wherein the weight of the pathological T component for predicting prognosis is upgraded as a dominant factor and the importance of nodal involvement depends on T category[5]. Tumor deposits (TDs) can upgrade stage across pT and pN categories in the 9^th edition. Here, we assessed the accuracy and clinical relevance of the 9^th edition AJCC staging system.

MATERIALS AND METHODS

Study population

Consecutive patients with pathologically confirmed primary CC and rectal cancer (RC) who underwent curative-intent surgical resection from July 2015 to July 2021 were enrolled. The inclusion criteria were: (1) Histologically verified malignancy following radical resection performed either laparoscopically or via open laparotomy; and (2) Availability of comprehensive clinicopathological and follow-up records. The exclusion criteria were: (1) Postoperative metastasis or death within 3 months; (2) Distant metastases upon initial diagnosis; (3) Follow-up duration < 2 years; and (4) Pathological T0 staging.

Baseline clinical and pathological information

Clinical and pathological data were extracted retrospectively from electronic medical records. Demographic and surgical details (age, sex, operative date, tumor location, and metastatic status) were recorded. Pathological assessments included nodal lymph nodes (positive/total lymph nodes), TDs, and pT stage. Cases with incomplete pathology data were excluded.

Patient outcomes

Recurrence and mortality data were obtained via telephonic interviews with patients/families or electronic record review. Disease-free survival (DFS) was estimated from the day of primary surgery to the date of diagnosis of recurrent disease or the last date of follow-up without recurrence or the date of death from other causes. OS was estimated from the day of primary surgery to death from any causes or the last day of follow-up for those who were alive.

Therapeutic protocol and surveillance

All patients underwent radical surgery via complete mesocolic excision for CC and total mesorectal excision for RC. Neoadjuvant chemotherapy or adjuvant chemotherapy with the CAPOX (capecitabine + oxaliplatin) or FOLFOX (fluorouracil + oxaliplatin + leucovorin) regimen was applied over six cycles, or neoadjuvant chemoradiotherapy or adjuvant chemoradiotherapy was recommended if necessary. Postoperative surveillance followed National Comprehensive Cancer Network guidelines: Quarterly visits for 2 years, semiannually for 3 subsequent years, and annually thereafter until 10 years. At every visit, interim clinical history and laboratory blood tests including carcinoembryonic antigen levels were checked.

Statistical analysis

Intergroup survival disparities (DFS/OS) were analyzed using Kaplan-Meier curves and log-rank tests. Significance was set at a two-sided P value < 0.05 with 95% confidence intervals. Analyses were conducted in R v4.0.5 (http://www.R-project.org).

RESULTS

Clinicopathological characteristics

During the study period, a total of 1251 patients with stage I-III CC and 1162 patients with stage I-III RC underwent curative surgery. As detailed in Table 1, the CC cohort comprised 533 female (42.6%) and 718 male (57.4%) patients, with a median age of 65 years. Pathological staging revealed the following T-stage distribution: PT1 in 40 cases (3.2%), pT2 in 132 cases (10.6%), pT3 in 991 cases (79.2%), pT4a in 32 cases (2.5%), and pT4b in 56 cases (4.5%). Lymph node involvement was distributed as follows: 774 patients (61.8%) had no positive lymph nodes (PLN), while 142 (11.4%), 117 (9.4%), 60 (4.8%), 44 (3.5%), and 114 (9.1%) had 1, 2, 3, 4, or > 4 PLNs, respectively. TDs were present in 186 cases (14.9%) and absent in 1065 (85.1%). For the RC cohort, 716 male (61.6%) and 446 female (38.4%) patients with RC were enrolled. Pathological staging revealed the following T-stage distribution: PT1 in 70 cases (6.0%), pT2 in 260 cases (22.4%), pT3 in 794 cases (68.3%) and pT4 in 38 cases (3.3%). Lymph node involvement was distributed as follows: 693 patients (59.6%) had no PLN, while 154 (13.3%), 89 (7.7%), 52 (4.5%), 36 (3.1%), and 138 (11.8%) had 1, 2, 3, 4, or > 4 PLNs, respectively.

Table 1 Clinicopathological characteristics of the colon cancer and rectal cancer cohorts, n (%).

	Overall (n = 2413)	Colon cancer (n = 1251)	Rectal cancer (n = 1162)
Sex
Male		533 (42.6)	716 (61.6)
Female		718 (57.4)	446 (38.4)
pT stage
T1		40 (3.2)	70 (6.0)
T2		132 (10.6)	260 (22.4)
T3		991 (79.2)	794 (68.3)
T4		88 (7.0)	38 (3.3)
Number of metastatic lymph nodes
0		774 (61.8)	693 (59.6)
1		142 (11.4)	154 (13.3)
2		117 (9.4)	89 (7.7)
3		60 (4.8)	52 (4.5)
4		44 (3.5)	36 (3.1)
> 4		114 (9.1)	138 (11.8)

Follow-up information

In the CC cohort, the median follow-up time for DFS and OS was 65.43 (48.90-82.47) months and 63.63 (46.53-81.77) months, respectively. In the RC cohort, the median follow-up time was 64.37 (48.07-83.93) months and 64.27 (46.87-83.17) months for DFS and OS, respectively.

Updated pathologic staging system

According to proposed changes to the pathologic staging for CC by the AJCCCCEP, CC with M0 disease was classified as three groups with only two subgroups within stage II, which is divided into three subgroups in the 8^th staging system. The distribution of pathological staging in the CC and RC cohort according to the 8^th and 9^th criteria is shown in Figure 1.

Figure 1 Distribution of pathological staging in the colon cancer and rectal cancer cohorts according to 8^th edition American Joint Committee on Cancer staging system and 9^th edition American Joint Committee on Cancer staging system. CC: Colon cancer; RC: Rectal cancer; AJCC: American Joint Committee on Cancer.

Kaplan-Meier survival curve analysis based on the 8^th edition and 9^th edition

The survival analysis of the CC cohort including stage description and the corresponding 5 years survival, as well as consecutive pair-wise comparison results are shown in Figure 2. The 5-year DFS based on the 9^th edition steadily decreased as stage number increased, and hazard ratio increased as stage number increased. Consistent with DFS, OS further substantiated the conclusion that the 9^th edition outperforms the previous system in terms of prognosis. Similarly, the same analytical approach was carried out on the RC cohort (Figure 3).

Figure 2 Kaplan-Meier survival curve analysis in the colon cancer cohort. A: Disease-free survival according to the 8^th edition American Joint Committee on Cancer (AJCC) staging system; B: Overall survival according to the 8^th edition AJCC staging system. C: Disease-free survival according to the 9^th edition AJCC staging system. D: Overall survival according to the 9^th edition AJCC staging system. DFS: Disease-free survival; OS: Overall survival; HR: Hazard ratio; TN: Primary Tumor and Regional Lymph Nodes.

Figure 3 Kaplan-Meier survival curve analysis in the rectal cancer cohort. A: Disease-free survival according to the 8^th edition American Joint Committee on Cancer (AJCC) staging system; B: Overall survival according to the 8^th edition AJCC staging system; C: Disease-free survival according to the 9^th edition AJCC staging system; D: Overall survival according to the 9^th edition AJCC staging system. DFS: Disease-free survival; OS: Overall survival; HR: Hazard ratio; TN: Primary Tumor and Regional Lymph Nodes.

Furthermore, the C-indices of the 8^th edition and 9^th edition were calculated in terms of DFS (0.689 and 0.674, P = 0.036) and OS (0.668 and 0.656, P = 0.180) in the CC cohort. Likewise, the predictive performance was also detected in cohort RC with C-indices of 0.685 vs 0.659 (P = 0.014) and 0.679 vs 0.662 (P = 0.168) in terms of DFS and OS.

Receiver operating characteristic curve based on the 8^th edition and 9^th edition

In the CC cohort, the area under the curves (AUC) at 1 year, 3 years and 5 years were 0.740, 0.726, 0.710 and 0.745, 0.734, 0.721 for the 8^th edition and 9^th edition system for DFS, respectively. For OS, the AUCs were 0.724, 0.720, 0.654 and 0.773, 0.738, 0.668 (Figure 4A-D). The integrated AUCs (iAUC) were 0.72 vs 0.731 (P = 0.06) and 0.704 vs 0.723 (P = 0.016) for the 8^th edition and 9^th edition in terms of DFS and OS, respectively. In the RC cohort, the AUCs at 1 year, 3 years and 5 years were 0.724, 0.733, 0.714 vs 0.737, 0.742, 0.722 in terms of DFS for the 8^th edition and 9^th edition, respectively, and 0.753, 0.726, 0.706 vs 0.774, 0.739, 0.711 for OS (Figure 4E-H). The iAUCs were 0.735 vs 0.744 (P = 0.096) and 0.741 vs 0.752 (P = 0.178) for the 8^th edition and 9^th edition in terms of DFS and OS, respectively. Time dependent receiver operating characteristic analysis revealed better performance of the 9^th edition in discriminatory power with a statistically significant difference for CC.

Figure 4 Receiver operating characteristic curve analysis. A: Receiver operating characteristic curve (ROC) analysis for disease-free survival (DFS) in the colon cancer (CC) cohort according to 8^th edition American Joint Committee on Cancer (AJCC) staging system; B: ROC analysis for overall survival (OS) in the CC cohort according to 8^th edition AJCC staging system; C: ROC analysis for DFS in the CC cohort according to 9^th edition AJCC staging system; D: ROC analysis for OS in the CC cohort according to 9^th edition AJCC staging system; E: ROC analysis for DFS in the rectal cancer (RC) cohort according to 8^th edition AJCC staging system; F: ROC analysis for OS in the RC cohort according to 8^th edition AJCC staging system; G: ROC analysis for DFS in the RC cohort according to 9^th edition AJCC staging system; H: ROC analysis for OS in the RC cohort according to 9^th edition AJCC staging system. CC: Colon cancer; RC: Rectal cancer; DFS: Disease-free survival; OS: Overall survival; AUC: Area under the curve.

DISCUSSION

In accordance with previous studies which indicated the nonhierarchical nature of current summary staging with respect to OS in patients with CC, in the present study, we demonstrated the prognostic value of the 9^th edition AJCC staging system, which exhibited superior discriminative performance than the 8^th edition AJCC staging system in terms of patient outcomes[6]. Different indicators for outcome including OS and DFS were assessed, respectively, and the survival curves were hierarchically separated among patients with different stages. The paradoxical phenomenon that patients with better survival but advanced staging observed in the current 8^th edition is attributed to the complicated interplay between T category and N category, while the current 8^th edition prioritized the N category[6,7].

According to the 9^th edition, patients were divided into six subgroups ranging from subgroup I featuring best survival to subgroup IIIc displaying poorest survival. Patients with stage I CC according to the 8^th edition show heterogeneous outcomes, which is reasonable as advanced T component is associated with poor outcome. Patients with IIIb CC exhibited the most various outcomes ranging from IIb to IIIc after restaging. Patients with IIa and most patients with IIIa, as well as a small portion of patients with IIIb shared similar outcome categorizing as IIb. A fraction of patients with IIIa and some with IIIb were classified as IIIa. Patients with IIb, IIc, a portion of IIIb and IIIc CC were grouped into IIIb according to the 9^th edition. Patients with the worst prognosis from stages IIIb and IIIc were aggregated into a single high-risk group designated as IIIc. The aforementioned findings are in agreement with previous studies, with Rottoli et al[8] showing that stage IIc cancers had poorer outcomes than IIIa and were comparable to IIIb cancers[8-11]. In addition, the finding in the current study that patients with IIIb CC had heterogeneous survival was also indicated in a previous study[12]. This stratification brings about novel insight into the tumor biological behaviors that the depth of tumor invasion into the colorectal wall calls for enhanced clinical vigilance. Moreover, the 9^th edition with homogeneous prognosis in specific groups may facilitate real effectiveness of therapy being evaluated in clinical trials.

Multiple studies have demonstrated that TDs are a factor for poor prognosis, in addition, in our previous study, we indicated that the presence of TDs warrants further investigation[13-15]. Another population-based study demonstrated that the concomitant presence of TDs and PLNs in CC was associated with worse OS and that TDs had similar prognostic implications to lymph node metastasis[16]. As an independent prognostic factor in CC, TDs are not well represented in the 8^th edition, wherein TDs are taken into consideration only when the lymph node is not involved[17]. Furthermore, the presence of TDs was associated with a worse prognosis across all pN categories, indicating that valuable prognostic information was lost by incorporating TDs into the N category[18]. In these new staging criteria, TDs upgrade stage across pT and pN categories, which provides more accuracy in the prediction of prognosis.

In this study, we included patients with CC as well as RC and the results suggest that the 9^th edition could also be useful for RC.

The current study has several limitations. First, the 9^th edition was established based on survival outcomes. Importantly, the conclusion should be carefully interpreted as the 9^th edition refined the 8^th edition staging with respect to competency of prognostic prediction, but does not confer prediction information regarding response to therapy or provide information on which treatment should optimally be used, which should follow the guideline derived from clinical trials. Second, patients receiving preoperative therapy lacked representation due to the small sample size and it remains uncertain if the 9^th edition can be extrapolated to these patients. Third, the current study did not take other prognostic characteristics of CC, such as peripheral neural invasion and pathological molecular profile into consideration, which limits the generalizability of findings[19-22].

CONCLUSION

The 9^th edition AJCC staging system can hierarchically separate survival outcome among patients with different disease stages, therefore provides better prognostic discrimination for stage I-III CC and RC than the 8^th edition. In addition, the 9^th edition AJCC staging system intuitively confers prognosis to patients and optimizes communication regarding prognosis to each patient’s healthcare team.