Content uploaded by Jing-Xu Sun
Author content
All content in this area was uploaded by Jing-Xu Sun on Aug 20, 2014
Content may be subject to copyright.
A preview of the PDF is not available
Is the prediction of prognosis not improved by the seventh edition of the TNM classification for colorectal cancer? Analysis of the surveillance, epidemiology, and end results (SEER) database
Abstract and Figures
Background
Whether the 7th edition of American Joint Committee on Cancer (AJCC) TNM staging system (AJCC-7) is a successful revision remains debatable. We aimed to compare the predictive capacity of the AJCC-7 for colorectal cancer with the 6th edition of the AJCC TNM staging system (AJCC-6).
Methods
The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) dataset consisting of 158,483 records was used in this study. We evaluated the predictive capacity of the two editions of the staging system using Harrell’s C index and Bayesian Information Criterion (BIC).
Results
There was a significant prognostic difference between patients at stage IIB and IIC (P < 0.001). Stage III patients with similar prognoses were adequately sub-grouped in the same stage according to AJCC-7. The Harrell’s C index revealed a value of 0.7692 for AJCC-7, which was significantly better than 0.7663 for AJCC-6 (P < 0.001). BIC analysis provided consistent results (P < 0.001).
Conclusions
This study demonstrates that AJCC-7 is superior to the AJCC-6 staging system in predictive capacity.
Figures - uploaded by Jing-Xu Sun
Author content
All figure content in this area was uploaded by Jing-Xu Sun
Content may be subject to copyright.
... Staging-related colorectal cancer prognosis studies have shown an inverse correlation between stage and prognosis, meaning that Stage I-II has a better prognosis than Stage III-IV. 2,[4][5][6] For patients with Stage I-II colorectal cancer, the most effective treatment is surgery, often without the need for adjuvant or neoadjuvant therapy. ...
Objective
This retrospective observational study aims to develop and validate artificial intelligence (AI) pathomics models based on pathological Hematoxylin–Eosin (HE) slides and pathological immunohistochemistry (Ki67) slides for predicting the pathological staging of colorectal cancer. The goal is to enable AI‐assisted accurate pathological staging, supporting healthcare professionals in making efficient and precise staging assessments.
Methods
This study included a total of 267 colorectal cancer patients (training cohort: n = 213; testing cohort: n = 54). Logistic regression algorithms were used to construct the models. The HE image features were used to build the HE model, the Ki67 image features were used for the Ki67 model, and the combined model included features from both the HE and Ki67 images, as well as tumor markers (CEA, CA724, CA125, and CA242). The predictive results of the HE model, Ki67 model, and tumor markers were visualized through a nomogram. The models were evaluated using ROC curve analysis, and their clinical value was estimated using decision curve analysis (DCA).
Results
A total of 260 deep learning features were extracted from HE or Ki67 images. The AUC for the HE model and Ki67 model in the training cohort was 0.885 and 0.890, and in the testing cohort, it was 0.703 and 0.767, respectively. The combined model and nomogram in the training cohort had AUC values of 0.907 and 0.926, and in the testing cohort, they had AUC values of 0.814 and 0.817. In clinical DCA, the net benefit of the Ki67 model was superior to the HE model. The combined model and nomogram showed significantly higher net benefits compared to the individual HE model or Ki67 model.
Conclusion
The combined model and nomogram, which integrate pathomics multi‐modal data and clinical‐pathological variables, demonstrated superior performance in distinguishing between Stage I–II and Stage III colorectal cancer. This provides valuable support for clinical decision‐making and may improve treatment strategies and patient prognosis. Furthermore, the use of immunohistochemistry (Ki67) slides for pathomics modeling outperformed HE slide, offering new insights for future pathomics research.
... Tumors that directly invade or adhere to adjacent organs or structures are considered T4b. 4 The incidence of the T4 colorectal tumors is around 5%-8.8% of the cases; among the advanced resected cases, the incidence of the T4 tumors is up to 21%-43% 5-7 using the SEER dataset between 1973 and 2008. Gao et al. 8 have found an incidence of 6.3% for pT4a and 4.4% for pT4b CRCs, but only in cases without distant metastasis. It is to be expected that in cases with distant metastasis, the incidence of the T4 CRCs is even higher. ...
The patient is a 95-year-old male with underlying disease of hypertensive cardiovascular disease, Type 2 diabetes mellitus, and old cerebrovascular accident. Persistent abdominal cramping pain with fullness sensation and poor appetite had been noted. The flexible colonoscopy could not pass through sigmoid colon. Abdomen computed tomography demonstrated tumor obstruction over sigmoid colon, measurable size in 5.2 cm, with highly suspicion of duodenum invasion. During operation, tumor which located in the sigmoid colon invaded to fourth portion of the duodenum. En bloc resection of the tumor, duodenorrhaphy, and feeding jejunostomy were performed. An end-to-end anastomosis and protective loop ileostomy formation at the right lower quadrant were performed as totally bypass of anastomoses consequently. Direct invasion of colorectal adenocarcinoma into adjacent structures occurs frequently, but only rarely in the duodenum involved. In the case presented above, we could aim that active surgical management is useful for improving patient prognosis without increasing the risk associated with surgery.
... In the 1950s, the TNM staging system (stage T, N, and M) was introduced to predict cancer patients' survivability [13]. Then, in 1994, Burke [14] used statistical techniques and proved that the performance of those techniques gives better results than the TNM staging system. ...
... Adjuvant chemotherapy is frequently used in stage III CRC, but it remains controversial for stage II disease. The 5-year disease specific survival for the T4 tumors is about 75.4% [5]; the observed 5-year survival rate for colon cancer in stage IIB (T4aN0M0) was 60.6%, significantly higher than 45.7% for stage IIC (T4bN0M0) [6]. Older patients are commonly under-represented in clinical trials and consequently, the standard therapeutic strategies are not fully validated in this population [7]. ...
Background: This study aimed to compare the outcomes of older and younger patients with T4 colorectal cancer (CRC) treated with surgery. Methods: Consecutive patients with T4 CRC treated surgically at Henri Mondor Hospital between 2008 and 2016 were retrospectively analyzed in age subgroups (1) 50–69 years and (2) ≥70 years for overall and relative survival. The multivariable analyses were adjusted for adjusted for age, margin status, lymph node involvement, CEA level, postoperative complications (POC), synchronous metastases, and type of surgery. Results: Of 106 patients with T4 CRC, 57 patients (53.8%) were 70 years or older. The baseline characteristics were generally balanced between the two age groups. Older patients underwent adjuvant therapy less commonly (42.9 vs. 57.1%; p = 0.006) and had a longer delay between surgery and chemotherapy (median 40 vs. 34 days; p < 0.001). A higher trend for POC was reported among the older patients but did not impact the survival outcomes. After adjusting for confounding factors, the overall survival was shorter among the older patients (HR = 3.322, 95% CI 1.49–7.39), but relative survival was not statistically correlated to the age group (HR = 0.873, 95% CI 0.383–1.992). Conclusions: Older patients with CRC were more prone to severe POC, but age did not impact the relative survival of patients with T4 colorectal cancer. Older patients should not be denied surgery based on age alone.
... When examining these results, it is important to underline that since 2002, three different American Joint Committee on Cancer (AJCC) staging classifications have been issued, with differences in stage grouping especially between 6 th and 7 th edition (17, 18). ...
Traditionally, lymph node metastases (LNM) evaluation is essential to the staging of colon cancer patients according to the TNM (tumor–node–metastasis) system. However, in recent years evidence has accumulated regarding the role of emerging pathological features, which could significantly impact the prognosis of colorectal cancer patients. Lymph Node Ratio (LNR) and Log Odds of Positive Lymph Nodes (LODDS) have been shown to predict patients’ prognosis more accurately than traditional nodal staging and it has been suggested that their implementation in existing classification could help stratify further patients with overlapping TNM stage. Tumor deposits (TD) are currently factored within the N1c category of the TNM classification in the absence of lymph node metastases. However, studies have shown that presence of TDs can affect patients’ survival regardless of LNM. Moreover, evidence suggest that presence of TDs should not be evaluated as dichotomic but rather as a quantitative variable. Extranodal extension (ENE) has been shown to correlate with presence of other adverse prognostic features and to impact survival of colorectal cancer patients. In this review we will describe current staging systems and prognostic/predictive factors in colorectal cancer and elaborate on available evidence supporting the implementation of LNR/LODDS, TDs and ENE evaluation in existing classification to improve prognosis estimation and patient selection for adjuvant treatment.
... Treatment regimens vary greatly among these stages, with local tumor excision at one end of the spectrum and neoadjuvant therapy with (procto)colectomy and adjuvant therapy at the other end. Although the TMN system provides a standardized guidance for management, outcomes vary depending on several factors [12,13]. For example, the expected benefit of adjuvant chemotherapy is higher in average-risk stage III patients compared to average-risk stage II patients. ...
Colorectal cancer is the third leading cause of cancer-related death, and its incidence
is rising in the younger patient population. In the past decade, research has unveiled several processes (underlying tumorigenesis, many of which involve interactions between tumor cells and the surrounding tissue or tumor microenvironment (TME). Interactions between components of the TME are mediated at a sub-microscopic level. However, the endpoint of those interactions results in morphologic changes which can be readily assessed at microscopic examination of biopsy and resection specimens. Among these morphologic changes, alteration to the tumor stroma is a new, important determinant of colorectal cancer progression. Different methodologies to estimate the proportion of tumor stroma relative to tumor cells, or tumor stroma ratio (TSR), have been developed.
Subsequent validation has supported the prognostic value, reproducibility and feasibility of TSR in various subgroups of colorectal cancer. In this manuscript, we review the literature surrounding TME in colorectal cancer, with a focus on tumor stroma ratio.
Colorectal carcinoma (CRC) is the second most common cancer and third leading cause of cancer-related deaths worldwide. Although the staging system provides a standardized guidance in treatment regimens, the clinical outcome in patients with colon cancer at the same TNM stage may vary dramatically. Thus, for better predictive accuracy, further prognostic and/or predictive markers are required. Patients who underwent curative surgery for colorectal cancer in past 3 years at a tertiary care hospital were retrospectively included in this cohort study to evaluate the prognostic indicators, tumor-stroma ratio (TSR) and tumor budding (TB) on histopathological sections and correlated them with pTNM staging, histopathological grading, tumor size, and lymphovascular and perineural invasion in patients with colo-rectal cancer. TB was strongly associated with advanced stage of the disease along with lympho-vascular and peri-neural invasion and it can be used as an independent adverse prognostic factor. TSR showed a better sensitivity, specificity, PPV and NPV as compared to TB in patients having poorly differentiated adenocarcinoma than those with moderately or well differentiated.
Despite the implementation of global screening programs, colorectal cancer (CRC) remains the second leading cause of cancer-related deaths worldwide. More than 10% of patients with colon cancer are diagnosed as having locally advanced disease with a relatively poor five-year survival rate. Locally advanced colon cancer (LACC) presents surgical challenges to R0 resection. The advantages and disadvantages of preoperative radiotherapy for LACC remain undetermined. Although several reliable novel biomarkers have been proposed for the prediction and prognosis of CRC, few studies have focused solely on the treatment of LACC. This comprehensive review highlights the role of predictive biomarkers for treatment and postoperative oncological outcomes for patients with LACC. Moreover, this review discusses emerging needs and approaches for the discovery of biomarkers that can facilitate the development of new therapeutic targets and surveillance of patients with LACC.
In colorectal cancer a T4 situation is present in cases of perforation of the visceral peritoneum (T4a) or penetration into an adjacent organ (T4b). The particular aspect of this tumor stage concerns the local conditions at the site of the primary tumor and it is therefore primarily a case for surgery. The usual standard examinations do not always give a complete image of T4 tumors and should be supplemented as required. Depending on the location and extent of involvement of adjacent organs or structures, a resection with curative intent can only be achieved by a multivisceral, often interdisciplinary concept. Emergency scenarios are frequent and challenging. In order to maintain a balanced relationship between morbidity and benefits in extensive interventions, a thorough assessment of patients and resources is essential.
Medical researchers frequently make statements that one model pre- dicts survival better than another, and they are frequently challenged to provide rigorous statistical justification for those statements. Stata provides the estat concordance command to calculate the rank parameters Harrell’s C and Somers’ D as measures of the ordinal predictive power of a model. However, no confidence limits or p-values are provided to compare the predictive power of distinct models. The somersd package, downloadable from Statistical Software Components, can provide such confidence intervals, but they should not be taken seriously if they are calculated in the dataset in which the model was fit. Methods are demonstrated for fitting alternative models to a training set of data, and then measuring and comparing their predictive powers by using out-of-sample prediction and somersd in a test set to produce statistically sensible confidence intervals and p-values for the differences between the predictive powers of different models. Copyright 2010 by StataCorp LP.
The American Joint Committee on Cancer's Cancer Staging Manual is used by physicians throughout the world to diagnose cancer and determine the extent to which cancer has progressed. All of the TNM staging information included in this Sixth Edition is uniform between the AJCC (American Joint Committee on Cancer) and the UICC (International Union Against Cancer).
In addition to the information found in the Handbook, the Manual provides standardized data forms for each anatomic site, which can be utilized as permanent patient records, enabling clinicians and cancer research scientists to maintain consistency in evaluating the efficacy of diagnosis and treatment. The CD-ROM packaged with each Manual contains printable copies of each of the book’s 45 Staging Forms.
Multivariable regression models are powerful tools that are used frequently in studies of clinical outcomes. These models can use a mixture of categorical and continuous variables and can handle partially observed (censored) responses. However, uncritical application of modelling techniques can result in models that poorly fit the dataset at hand, or, even more likely, inaccurately predict outcomes on new subjects. One must know how to measure qualities of a model's fit in order to avoid poorly fitted or overfitted models. Measurement of predictive accuracy can be difficult for survival time data in the presence of censoring. We discuss an easily interpretable index of predictive discrimination as well as methods for assessing calibration of predicted survival probabilities. Both types of predictive accuracy should be unbiasedly validated using bootstrapping or cross-validation, before using predictions in a new data series. We discuss some of the hazards of poorly fitted and overfitted regression models and present one modelling strategy that avoids many of the problems discussed. The methods described are applicable to all regression models, but are particularly needed for binary, ordinal, and time-to-event outcomes. Methods are illustrated with a survival analysis in prostate cancer using Cox regression.
4020 Background: The 6th edition (ed) of AJCC staging subdivided Stage II into IIA(T3N0)/IIB(T4N0) and Stage III into IIIA(T1–2N1M0), IIIB(T3–4N1M0), IIIC(anyTN2M0). Subsequent analyses supported further substaging of Stage III due to improved survival with T1–2N2 cancers. Before making such changes in the AJCC 7th ed, the HTF sought validation in a population-based dataset that depth of invasion interacts with nodal status to affect survival. Methods: SEER population-based data from Jan 1, 1992 to Dec 2004 for 35,829 rectal and 109,953 colon cancer pts was compared to Cooperative Group rectal data on 2551 pts (Rectal Pooled Analysis). Tumors were stratified by SEER’s ‘extent of disease’ and ‘number of positive nodes’ coding schemes. T4N0 cancers were stratified by ‘tumors that perforate visceral peritoneum’ vs ‘tumors that invade or are adherent to adjacent organs or structures’. N1 and N2 were stratified by number of N+: N1(1 vs 2–3), N2(4–6 vs >7). Observed 5-y survival (∼ overall survival, OS) was obt...
Multivariable regression models are powerful tools that are used frequently in studies of clinical outcomes. These models can use a mixture of categorical and continuous variables and can handle partially observed (censored) responses. However, uncritical application of modelling techniques can result in models that poorly fit the dataset at hand, or, even more likely, inaccurately predict outcomes on new subjects. One must know how to measure qualities of a model's fit in order to avoid poorly fitted or overfitted models. Measurement of predictive accuracy can be difficult for survival time data in the presence of censoring. We discuss an easily interpretable index of predictive discrimination as well as methods for assessing calibration of predicted survival probabilities. Both types of predictive accuracy should be unbiasedly validated using bootstrapping or cross-validation, before using predictions in a new data series. We discuss some of the hazards of poorly fitted and overfitted regression models and present one modelling strategy that avoids many of the problems discussed. The methods described are applicable to all regression models, but are particularly needed for binary, ordinal, and time-to-event outcomes. Methods are illustrated with a survival analysis in prostate cancer using Cox regression.
The seventh edition of the American Joint Committee on Cancer (AJCC) staging system has new substages for colon cancer. We used survival data from a single medical center to analyze this new AJCC edition.
The colon cancer database of Taipei Veterans General Hospital provided 1,865 patient records covering from 1999 to 2005. Survival rates were evaluated using the Kaplan-Meier method.
There were 268, 607, 561, and 421 patients in stages I, II, III, and IV disease with 5-year observed survival rates of 86.3%, 79.2%, 65.4%, and 12.8%, respectively. Survival rates were not significantly different between those with T4a and T4b disease (P = 0.806). The outcome of N1c disease was similar to N1a and N1b but worse than N0 (P = 0.004). Survival rates for M1a and M1b disease became different after reclassifying solely peritoneal seeding as M1a (P < 0.001). No discrepancy of outcomes between stage IIIA and stage IIB/IIC remained in the seventh edition.
Evolution from the fifth to seventh edition of the AJCC staging system is successful in separating prognostic groups by substaging. But some issues remain unresolved, including the subdivision of T4, N1, and M1.
To compare the prognostic value of the sixth and seventh editions of the TNM classification, and of additional prognostic factors, in colorectal cancer.
The seventh TNM edition was released in 2009 with the aim of providing a more precise prediction of prognosis.
Clinical and histopathological data of 2229 patients with colorectal cancer who underwent tumor resection between 1990 and 2006 were analyzed and compared by using the sixth and seventh editions of the TNM classification and a statistically calculated model of prognostic factors.
With the sixth edition, 5-year survival was 96% for stage I, 90% for IIA, 86% for IIB, 90% for IIIA, 72% for IIIB, 48% for IIIC, and 13% for IV. With the seventh edition, 5-year survival was 96% for stage I, 90% for IIA, 84% for IIB, 87% for IIC, 89% for IIIA, 72% for IIIB, 36% for IIIC, 15% for IVA, and 10% for IVB. The stage shifted for only 155 (7%) patients: from IIB to IIC (2%), from IIIB to IIIC (1%), and from IIIC to IIIA/B (4%). The performance of the seventh edition [concordance index (c-index) 0.83; 95% confidence interval (CI), 0.82-0.85] revealed no relevant improvement compared with the sixth edition (c-index 0.83; 95% CI, 0.82-0.84), or compared to a model based on independent prognostic factors (c-index 0.84; 95% CI, 0.83-0.86).
The seventh TNM edition did not provide greater accuracy in predicting colorectal cancer patients' prognosis but resulted in a more complex classification for daily clinical use.
The sixth and seventh editions of the American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) system for patients with stage II and stage III colorectal carcinoma (AJCC-6 and AJCC-7) were compared.
Between 2000 and 2007, 2511 stage II/III colorectal carcinoma patients received primary surgical resection at the Asan Medical Center (Seoul, Korea). All patients were staged using AJCC-6 and AJCC-7 TNM systems. Patients with synchronous or other cancers, those given preoperative chemotherapy or radiotherapy and those in whom fewer than 12 lymph nodes were resected, were excluded. Overall survival (OS) and disease-free survival (DFS) were compared.
Of 2511 patients, 255 (10.2%) had different stages in the AJCC-6 and AJCC-7. For the AJCC-7, the 5-year OS by stage was 94.2% for stage IIA, 88.8% for stage IIB, 83.5% for stage IIC, 91.8% for stage IIIA, 81.8% for stage IIIB and 72.0% for stage IIIC. The OS and the DFS were not significantly different for the new substages IIB (n = 57) and IIC (n = 34) (P = 0.34 and P = 0.87, respectively). For the 187 patients with stage T3N2a cancer, the OS and the DFS were significantly different from stage IIIB other than T3N2a (P = 0.008 and P = 0.01, respectively) and there were no statistically significant differences in OS between the T3N2a group and the IIIC group (P = 0.46).
The study indicates that AJCC-7 has better prognostic validity than AJCC-6 for staging of patients with stage II and stage III colorectal carcinoma.
Each year, the American Cancer Society estimates the numbers of new cancer cases and deaths expected in the United States in the current year and compiles the most recent data on cancer incidence, mortality, and survival based on incidence data from the National Cancer Institute, the Centers for Disease Control and Prevention, and the North American Association of Central Cancer Registries and mortality data from the National Center for Health Statistics. A total of 1,596,670 new cancer cases and 571,950 deaths from cancer are projected to occur in the United States in 2011. Overall cancer incidence rates were stable in men in the most recent time period after decreasing by 1.9% per year from 2001 to 2005; in women, incidence rates have been declining by 0.6% annually since 1998. Overall cancer death rates decreased in all racial/ethnic groups in both men and women from 1998 through 2007, with the exception of American Indian/Alaska Native women, in whom rates were stable. African American and Hispanic men showed the largest annual decreases in cancer death rates during this time period (2.6% and 2.5%, respectively). Lung cancer death rates showed a significant decline in women after continuously increasing since the 1930s. The reduction in the overall cancer death rates since 1990 in men and 1991 in women translates to the avoidance of about 898,000 deaths from cancer. However, this progress has not benefitted all segments of the population equally; cancer death rates for individuals with the least education are more than twice those of the most educated. The elimination of educational and racial disparities could potentially have avoided about 37% (60,370) of the premature cancer deaths among individuals aged 25 to 64 years in 2007 alone. Further progress can be accelerated by applying existing cancer control knowledge across all segments of the population with an emphasis on those groups in the lowest socioeconomic bracket.