Abstract
Background
The role of horizontal growth index of tumor size in survival prediction is still underappreciated in colon cancer because of the identification of vertical infiltration index reflected by T stage. We sought to reveal the impact of T stage on the prognostic and predictive value of tumor size in colon cancer.
Materials and Methods
Data of patients with stage I–III colon cancer were extracted from Surveillance, Epidemiology, and End Results Program (SEER) and Fudan University Shanghai Cancer Center (FUSCC) databases. Harrell's concordance index (c‐index) and time‐dependent receiver operating characteristic curve (ROC) were used to analyze the discriminative ability of prognostic factors.
Results
Stratified analyses based on T stage found that the increase of T stage significantly and negatively repressed the effect of tumor size on death and recurrence risk. In addition, tumor size showed the greatest hazard ratio of cancer‐specific death and relapse in T1 colon cancer. Even more importantly, the discriminatory ability of tumor size outperformed any other widely accepted prognostic clinical features in predicting cancer‐specific survival (SEER: c‐index 0.637, area under the ROC [AUC] 0.649; FUSCC: c‐index 0.673, AUC 0.686) and disease‐free survival (FUSCC: c‐index 0.645, AUC 0.656) in T1 stage colon cancer.
Conclusion
Tumor size is a critical clinical factor with considerable prognostic and predictive value for T1 colon cancer, and it should be selectively incorporated into the current staging system to facilitate prediction of death and recurrence risk.
Implications for Practice
To date, no consensus has been reached about the prognostic and predictive value of tumor size in colon cancer. Although tumor size is an independent prognostic factor for patients with colon cancer, the impact of tumor size on death or recurrence risk decreased notably with the increase of T stage. More importantly, the discriminative ability of tumor size outperformed any other clinical factors including N stage in patients with T1 colon cancer. Therefore, tumor size should be recommended to be incorporated into current staging systems to facilitate prognosis prediction for patients with T1 colon cancer.
Keywords: T1 stage, Colon cancer, Tumor size, Cancer‐specific survival, Disease‐free survival
Short abstract
Large horizontal tumor extent is considered to be associated with a negative prognosis. This article analyzes the prognostic and predictive value of tumor size in colon cancer by T stage.
Introduction
Colon cancer is one of the most common malignant tumors in the world 1. With the development of therapeutic regimens, the past decades have witnessed a significant improvement in the prognosis of patients with colon cancer. Currently, tumor‐node‐metastasis (TNM) staging is the most widely accepted system for risk stratification in colon cancer 2. However, the prognosis of patients within the same TNM stage varies strikingly. Consequently, more prognostic factors easily obtained from daily medical records should be identified to improve prognosis prediction and to individualize treatment strategy for patients with colon cancer.
Tumor size, defined as the maximal horizontal tumor diameter, has been studied for decades, and large horizontal tumor extent has generally been considered to be negatively associated with prognosis in many solid tumors, including colonic malignancies 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13. However, because of the vertical index of infiltration depth reflected by T stage, tumor size has not been recognized as a valuable factor in predicting prognosis in colon cancer. Indeed, T stage is a robust risk factor in colon cancer, but no previous studies have clarified the impact of T stage on the independent prognostic role and predictive ability of tumor size. Moreover, as an important horizontal tumor growth index, tumor size may have a unique role in specific subgroup of colon cancer with certain infiltration depth.
In this study, we analyzed the prognostic value of tumor size and compared its prognostic accuracy with other widely accepted factors in colon cancer with different T stage using data extracted from the Surveillance, Epidemiology, and End Results (SEER)–registered database. Moreover, because SEER data lack information about complete preoperative treatment and quality of operation, we further validated the results from SEER database in Fudan University Shanghai Cancer Center Database (FUSCC).
Subjects, Materials, and Methods
Patient Selection in the SEER Database
Patient data were obtained from the SEER database (http://seer.cancer.gov/seerstat), which is sponsored by the U.S. National Cancer Institute. It currently collects and publishes cancer incidence and survival data covering 18 population‐based cancer registries that represent approximately 28% of the population in the U.S. Using the SEER database, we identified a total of 121,926 primary colon cancers diagnosed between 2010 and 2014. Patients who met the following criteria were included: (A) pathologically diagnosed colon cancer; (B) patients received major surgical resection; (C) histological types were limited to adenocarcinoma (8140/3, 8210/3, 8261/3, 8263/3), mucinous adenocarcinoma (8480/3), and signet‐ring cell carcinoma (8490/3); (D) American Joint Committee on Cancer (AJCC) stage I–III; and (E) colon cancer was the only primary tumor. Patients with unknown information of detailed AJCC stage, tumor size, and follow‐up were excluded. Finally, a total of 52,513 patients were identified from SEER database in this study.
Patient Selection in the FUSCC Database
Patients with colon cancer diagnosed from 2006 to 2015 in FUSCC were identified in the present study. The inclusion criteria were as follows: (A) pathologically diagnosed colon cancer; (B) patients received radical surgical resection; (C) histological types were limited to adenocarcinoma (8140/3, 8210/3, 8261/3, 8263/3), mucinous adenocarcinoma (8480/3), and signet‐ring cell carcinoma (8490/3); (D) AJCC stage I–III; (E) colon cancer was the only primary tumor; and (F) patients were not treated with preoperative chemotherapy or endoscopic resection. After exclusion, a total of 4,398 patients with colon cancer were identified. The study was approved by the Ethical Committee and Institutional Review Board of the FUSCC, and written informed consent was signed by all the patients in FUSCC.
Statistical Analysis
Demographic and clinicopathological variables of age, gender, grade, tumor site, tumor size, T and N stage, adjuvant therapy regimens, and regional lymph node harvest (LNH) were retrieved from the SEER database. Furthermore, preoperative carcinoembryonic antigen (CEA) level, peripheral nerve invasion, and lymph‐vascular invasion were also extracted from FUSCC database. All included patients were restaged based on the AJCC Cancer Staging Manual (8th edition). Tumor size was analyzed as continuous variable in this study. Cancer‐specific survival (CSS) was the primary endpoint for the analysis based on both SEER and FUSCC database. As disease‐free survival (DFS) was available in FUSCC, DFS was also set as another primary endpoint. DFS was defined as the time from surgical treatment to first recurrence or end of life, and CSS represented the time period from treatment to death from colon cancer. The survival data of patients from FUSCC cohort were provided by the Clinical Statistics Center of FUSCC, relying on the hospital medical records follow‐up platform or contacts with patients by phone or e‐mail. Patients whose death or relapse status information during the follow‐up period was not available and patients who were still alive or absent any relapse at last follow‐up were censored for survival analysis.
Univariate analysis was utilized to examine the parameters that are significantly associated with CSS and DFS. A multivariate Cox proportional hazards model was built to verify the independent role of prognostic factors. The discriminative ability of tumor size and other factors were evaluated by using the Harrell's concordance index (c‐index) and time‐dependent receiver operating characteristic curve (ROC) 14, 15. A predictive variable with a higher c statistic and area under the ROC (AUC) indicated a better discrimination ability or prognostic accuracy. AUC at 5 years was analyzed and used in this study. Two‐sided p < .05 was considered statistically significant. All of the statistical analyses were performed with R (version 3.2.5, http://www.r-project.org).
Results
Patient Characteristics
Among 52,513 patients identified from the SEER database, 16,598 (31.6%) patients were aged less than 60 years. Of the patients, 27,033 (51.5%) were female, and 25,480 (48.5%) were male. Diagnoses were as follows: 27,286 (52.0%) patients were diagnosed with right‐side colon cancer, 5,218 (9.9%) with transverse colon cancer, and 20,009 (38.1%) with left‐side colon cancer. Grading was as follows: 4,559 (8.7%) patients had grade I, 37,556(71.5%) were grade II, and 9,390 (17.9%) were grade III or IV colon cancer. Distributions of colon cancer in adenocarcinoma, mucinous adenocarcinoma, and signet‐ring cell carcinoma were 47,529 (90.5%), 4,432 (8.4 %), and 552 (1.1%), respectively. The proportions of N0, N1a, N1b, N1c, N2a, and N2b stage were 61.8%, 11.6%, 11.5%, 2.4%, 7%, and 5.7%, respectively. The median (interquartile range) of tumor size was 4.1 (2.8–6.0) cm. Similar distribution of features was observed in patients from the FUSCC database. However, the incidence of male patients was notably higher in the FUSCC database. Detailed information about clinical pathological features is shown in Table 1.
Table 1.
Baseline clinicopathological features of patients with colon cancer stratified by T stage in SEER and FUSCC data sets
| Features | SEER, n (%) | FUSCC, n (%) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| All | T1 | T2 | T3 | T4 | All | T1 | T2 | T3 | T4 | |
| Age, years | ||||||||||
| ≤60 | 16,598 (31.6) | 2,404 (34.89) | 2,424 (28.38) | 9,157 (31.27) | 2,613 (33.49) | 1,921 (43.68) | 109 (49.19) | 214 (47.63) | 883 (42.33) | 715 (43.56) |
| >60 | 35,915 (68.4) | 4,487 (65.11) | 6,116 (71.62) | 20,122 (68.73) | 5,190 (66.51) | 2,477 (56.32) | 113 (50.81) | 235 (52.37) | 1,203 (57.67) | 926 (56.44) |
| Sex | ||||||||||
| Female | 27,033 (51.5) | 3,347 (48.57) | 4,495 (52.63) | 15,052 (51.41) | 4,139 (53.04) | 1,736 (39.48) | 104 (46.68) | 212 (47.13) | 810 (38.82) | 611 (37.24) |
| Male | 25,480 (48.5) | 3,544 (51.43) | 4,045 (47.37) | 14,227 (48.59) | 3,664 (46.96) | 2,662 (60.52) | 118 (53.32) | 237 (52.87) | 1,276 (61.18) | 1,030 (62.76) |
| Location | ||||||||||
| Right | 27,286 (52) | 3,062 (44.43) | 4,780 (55.97) | 15,288 (52.21) | 4,156 (53.26) | 2,101 (47.8) | 88 (39.6) | 175 (39) | 1,069 (51.2) | 769 (46.9) |
| Transverse | 5,218 (9.9) | 605 (8.78) | 730 (8.55) | 3,136 (10.71) | 747 (9.57) | 213 (4.8) | 9 (4.1) | 22 (4.9) | 97 (4.7) | 85 (5.2) |
| Left | 20,009 (38.1) | 3,224 (46.79) | 3,030 (35.48) | 10,855 (37.07) | 2,900 (37.17) | 2,084 (47.4) | 125 (56.3) | 252 (56.1) | 920 (44.1) | 787 (48) |
| Grade | ||||||||||
| I | 4,559 (8.7) | 1,464 (21.25) | 862 (10.09) | 1,823 (6.23) | 410 (5.25) | 338 (7.7) | 92 (41.4) | 73 (16.3) | 110 (5.3) | 63 (3.8) |
| II | 37,556 (71.5) | 4,600 (66.75) | 6,677 (78.19) | 21,476 (73.35) | 4,803 (61.55) | 3,572 (81.2) | 114 (51.4) | 352 (78.4) | 1,759 (84.3) | 1,347 (82.1) |
| III/IV | 9,390 (17.9) | 451 (6.54) | 885 (10.36) | 5,608 (19.15) | 2,446 (31.35) | 357 (8.1) | 3 (1.4) | 13 (2.9) | 167 (8) | 174 (10.6) |
| Unknown | 1,008 (1.9) | 376 (5.46) | 116 (1.36) | 372 (1.27) | 144 (1.85) | 131 (3) | 13 (5.9) | 11 (2.4) | 50 (2.4) | 57 (3.5) |
| Histology | ||||||||||
| AC | 47,529 (90.5) | 6,703 (97.27) | 7,958 (93.19) | 26,307 (89.85) | 6,561 (84.08) | 3,636 (82.7) | 218 (98.2) | 409 (91.1) | 1,696 (81.3) | 1,313 (80) |
| MAC | 4,432 (8.4) | 170 (2.47) | 552 (6.46) | 2,691 (9.19) | 1,019 (13.06) | 679 (15.4) | 4 (1.8) | 39 (8.7) | 348 (16.7) | 288 (17.6) |
| SRCC | 552 (1.1) | 18 (0.26) | 30 (0.35) | 281 (0.96) | 223 (2.86) | 83 (1.9) | 0 (0) | 1 (0.2) | 42 (2) | 40 (2.4) |
| LNH | ||||||||||
| <12 | 7,833 (14.9) | 2,328 (33.78) | 1,336 (15.64) | 3,176 (10.85) | 993 (12.73) | 443 (10.1) | 65 (29.3) | 68 (15.1) | 121 (5.8) | 189 (11.5) |
| ≥12 | 44,680 (85.1) | 4,563 (66.22) | 7,204 (84.36) | 26,103 (89.15) | 6,810 (87.27) | 3,955 (89.9) | 157 (70.7) | 381 (84.9) | 1,965 (94.2) | 1,452 (88.5) |
| N stage | ||||||||||
| N0 | 32,431 (61.8) | 6,196 (89.91) | 6,876 (80.52) | 16,297 (55.66) | 3,062 (39.24) | 2,462 (56) | 200 (90.1) | 350 (78) | 1,180 (56.6) | 732 (44.6) |
| N1a | 6,079 (11.6) | 385 (5.59) | 768 (8.99) | 3,814 (13.03) | 1,112 (14.25) | 531 (12.1) | 14 (6.3) | 51 (11.4) | 258 (12.4) | 208 (12.7) |
| N1b | 6,062 (11.5) | 217 (3.15) | 528 (6.18) | 4,086 (13.96) | 1,231 (15.78) | 620 (14.1) | 5 (2.3) | 26 (5.8) | 300 (14.4) | 289 (17.6) |
| N1c | 1,282 (2.4) | 28 (0.41) | 84 (0.98) | 837 (2.86) | 333 (4.27) | 355 (8.1) | 2 (0.9) | 8 (1.8) | 157 (7.5) | 188 (11.5) |
| N2a | 3,680 (7) | 54 (0.78) | 209 (2.45) | 2,439 (8.33) | 978 (12.53) | 293 (6.7) | 1 (0.5) | 8 (1.8) | 127 (6.1) | 157 (9.6) |
| 2b | 2,979 (5.7) | 11 (0.16) | 75 (0.88) | 1,806 (6.17) | 1,087 (13.93) | 137 (3.1) | 0 (0) | 6 (1.3) | 64 (3.1) | 67 (4.1) |
| CEA level | ||||||||||
| Low | NA | NA | NA | NA | NA | 2,799 (63.6) | 196 (88.7) | 368 (82) | 1,282 (61.5) | 953 (58.1) |
| High | NA | NA | NA | NA | NA | 1,599 (36.4) | 25 (11.3) | 81 (18) | 804 (38.5) | 688 (41.9) |
| LVI | ||||||||||
| No | NA | NA | NA | NA | NA | 3,300 (75) | 204 (92.3) | 389 (86.6) | 1,556 (74.6) | 1,151 (70.1) |
| Yes | NA | NA | NA | NA | NA | 1,098 (25) | 17 (7.7) | 60 (13.4) | 530 (25.4) | 490 (29.9) |
| PNI | ||||||||||
| No | NA | NA | NA | NA | NA | 3,455 (78.6) | 215 (97.3) | 431 (96) | 1,582 (75.8) | 1,227 (74.8) |
| Yes | NA | NA | NA | NA | NA | 943 (21.4) | 6 (2.7) | 18 (4) | 504 (24.2) | 414 (25.2) |
|
Tumor size, median (IQR) |
4.1 (2.8–6.0) | 1.4 (0.7–2.4) | 3.3 (2.4–4.5) | 4.6 (3.5–6.1) | 5.5 (4.0–7.5) | 4.0 (3.0–6.0) | 2.0 (1.5–3.0) | 3.5 (2.5–4.5) | 4.5 (3.5–6.0) | 5.0 (3.5–6.5) |
Abbreviations: AC, adenocarcinoma; CEA, carcinoembryonic antigen; FUSCC, Fudan University Shanghai Cancer Center; IQR, interquartile range; LNH, lymph node harvest; LVI, lymph‐vascular invasion; MAC, mucinous adenocarcinoma; NA, not available; PNI, peripheral nerve invasion; SEER, Surveillance, Epidemiology, and End Results; SRCC, signet‐ring cell carcinoma.
Impact of T Stage on the Prognostic Value of Tumor Size
Univariate and multivariate Cox regression models were used to evaluate the prognostic value of the tumor size. The results of the Cox regression analysis conducted in all the patients and stratified by T stage are illustrated in Figure 1. In all the patients, tumor size was an independent prognostic factor both in the SEER and FUSCC data sets. Large tumor size was negatively associated with death and recurrence risk (Fig. 1).
Figure 1.
Impact of T stage on the hazard ratio of tumor size in predicting CSS and DFS of colon cancer. (A): CSS in SEER database. (B): CSS in FUSCC database. (C): DFS in FUSCC database.
Abbreviations: CSS, cancer‐specific survival; DFS, disease‐free survival; FUSCC, Fudan University Shanghai Cancer Center; SEER, Surveillance, Epidemiology, and End Results.
In the SEER data set, subgroup analysis based on T stage found that the independent prognostic role of tumor size was not influenced by T stage, but its impact on death risk reflected by hazard ratio (HR) decreased strikingly with the increment of T stage (Fig. 1A). In the FUSCC cohort, T stage did not influence the independent value of tumor size in predicting CSS either (Fig. 1B). However, tumor size was no longer a prognosis predictor in T4 colon cancer (Fig. 1C). Similar to the results from the SEER data set, the impact of tumor size on both death and relapse risk declined dramatically with increase of infiltration depth (Fig. 1B, C). In summary, tumor size was associated with the greatest impact on death and relapse risk in T1 colon cancer, whereas the impact of tumor size on death or relapse risk decreased aberrantly in T4 colon cancer.
Discriminatory Ability of Tumor Size Outperformed Any Other Clinical Factors in T1 Stage
C‐index and time‐dependent ROC were used to evaluate the predictive ability of clinical factors in the SEER and FUSCC data sets. Overall, among all the patients with stage I–III colon cancer, N stage is expectedly the best predictor for CSS (SEER: c‐index 0.667, AUC 0.689; FUSCC: c‐index 0.686, AUC 0.637) and DFS (c‐index 0.677, AUC 0.642) with the highest c‐index and AUC (Table 2). Compared with N stage, tumor size did perform more poorly in predicting CSS (SEER: c‐index 0.618, AUC 0.608; FUSCC: c‐index 0.563, AUC 0.527) and DFS (c‐index 0.548, AUC 0.511; Table 2). Further subgroup analyses based on T stage were conducted, and we found that in patients with stage T1 colon cancer, tumor size outperformed any other factors in predicting CSS (SEER: c‐index 0.637, AUC 0.649; FUSCC: c‐index 0.673, AUC 0.686) and DFS (c‐index 0.645, AUC 0.656) with a considerable predictive ability, whereas N stage failed to be a good predictor for CSS (SEER: c‐index 0.525, AUC 0.563; FUSCC: c‐index 0.599, AUC 0.618) and DFS (c‐index 0.531, AUC 0.601). However, once T stage increased over T1 stage, the predictive ability of tumor size reduced dramatically and even became negligible (Table 2). In contrast, the predictive ability of N stage increased significantly after T stage became more advanced than T1 (Table 2).
Table 2.
Discriminatory ability of clinicopathological factors in predicting survival in colon cancer
| Variables | All | T1 | T2 | T3 | T4 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| C‐index | AUC | C‐index | AUC | C‐index | AUC | C‐index | AUC | C‐index | AUC | |
| SEER‐CSS | ||||||||||
| Tumor size | 0.618 | 0.608 | 0.637 | 0.649 | 0.535 | 0.534 | 0.535 | 0.509 | 0.532 | 0.489 |
| N stage | 0.667 | 0.689 | 0.525 | 0.563 | 0.584 | 0.605 | 0.63 | 0.651 | 0.615 | 0.649 |
| Grade | 0.588 | 0.574 | 0.551 | 0.542 | 0.538 | 0.534 | 0.549 | 0.536 | 0.583 | 0.569 |
| Histology | 0.523 | 0.521 | 0.51 | 0.506 | 0.503 | 0.493 | 0.511 | 0.508 | 0.512 | 0.508 |
| Location | 0.533 | 0.513 | 0.529 | 0.534 | 0.542 | 0.52 | 0.524 | 0.502 | 0.539 | 0.513 |
| LNH | 0.52 | 0.512 | 0.549 | 0.51 | 0.547 | 0.544 | 0.534 | 0.53 | 0.536 | 0.528 |
| FUSCC‐CSS | ||||||||||
| Tumor size | 0.563 | 0.527 | 0.673 | 0.686 | 0.521 | 0.569 | 0.548 | 0.509 | 0.533 | 0.487 |
| N stage | 0.686 | 0.637 | 0.599 | 0.618 | 0.578 | 0.581 | 0.666 | 0.645 | 0.678 | 0.635 |
| Grade | 0.581 | 0.556 | 0.623 | 0.466 | 0.628 | 0.59 | 0.55 | 0.556 | 0.572 | 0.553 |
| Histology | 0.543 | 0.529 | 0.508 | 0.489 | 0.54 | 0.531 | 0.515 | 0.506 | 0.544 | 0.53 |
| Location | 0.549 | 0.533 | 0.546 | 0.488 | 0.694 | 0.515 | 0.571 | 0.505 | 0.549 | 0.538 |
| LNH | 0.537 | 0.523 | 0.492 | 0.455 | 0.574 | 0.542 | 0.516 | 0.514 | 0.549 | 0.539 |
| CEA | 0.593 | 0.589 | 0.633 | 0.493 | 0.623 | 0.627 | 0.608 | 0.589 | 0.562 | 0.563 |
| LVI | 0.611 | 0.593 | 0.622 | 0.548 | 0.562 | 0.426 | 0.574 | 0.539 | 0.619 | 0.609 |
| PNI | 0.569 | 0.576 | 0.579 | 0.51 | 0.504 | 0.502 | 0.585 | 0.599 | 0.548 | 0.554 |
| FUSCC‐DFS | ||||||||||
| Tumor size | 0.548 | 0.511 | 0.645 | 0.656 | 0.539 | 0.533 | 0.551 | 0.544 | 0.507 | 0.463 |
| N stage | 0.677 | 0.642 | 0.531 | 0.601 | 0.576 | 0.527 | 0.699 | 0.639 | 0.645 | 0.628 |
| Grade | 0.567 | 0.548 | 0.59 | 0.333 | 0.594 | 0.607 | 0.569 | 0.552 | 0.549 | 0.539 |
| Histology | 0.53 | 0.518 | 0.508 | 0.489 | 0.561 | 0.545 | 0.527 | 0.506 | 0.52 | 0.512 |
| Location | 0.521 | 0.521 | 0.53 | 0.51 | 0.534 | 0.478 | 0.533 | 0.523 | 0.516 | 0.519 |
| LNH | 0.529 | 0.523 | 0.613 | 0.506 | 0.555 | 0.551 | 0.504 | 0.512 | 0.539 | 0.537 |
| CEA | 0.604 | 0.589 | 0.58 | 0.518 | 0.634 | 0.637 | 0.645 | 0.626 | 0.564 | 0.555 |
| LVI | 0.595 | 0.595 | 0.525 | 0.539 | 0.543 | 0.451 | 0.565 | 0.539 | 0.604 | 0.611 |
| PNI | 0.579 | 0.591 | 0.528 | 0.502 | 0.519 | 0.477 | 0.59 | 0.578 | 0.562 | 0.581 |
Abbreviations: AUC, area under the receiver operating characteristic curve; CEA, carcinoembryonic antigen; CSS, cancer‐specific survival; DFS, disease‐free survival; FUSCC, Fudan University Shanghai Cancer Center; LNH, lymph node harvest; LVI, lymph‐vascular invasion; PNI, peripheral nerve invasion; SEER, Surveillance, Epidemiology, and End Results.
To further clarify the predictive value of tumor size in T1 colon cancer, we stratified patients with stage T1 colon cancer into two groups based on lymph node status in SEER database. It is suggested that tumor size performed best among all the factors regardless of lymph node status and maintained its good discriminative ability in both groups (Table 3). This further stratified analysis was not conducted in the FUSCC cohort because of sample size limitation.
Table 3.
Discriminative power of prognostic factors in T1 colon cancer from the Surveillance, Epidemiology, and End Results database
| Variables | T1N0 stage | T1N+ stage | ||
|---|---|---|---|---|
| C‐index | AUC | C‐index | AUC | |
| Tumor size | 0.626 | 0.648 | 0.682 | 0.625 |
| N stage | 0.5 | 0.5 | 0.541 | 0.571 |
| Grade | 0.532 | 0.531 | 0.634 | 0.56 |
| Histology | 0.507 | 0.497 | 0.522 | 0.527 |
| Location | 0.531 | 0.532 | 0.599 | 0.577 |
| LNH | 0.545 | 0.511 | 0.603 | 0.558 |
Abbreviations: AUC, area under the receiver operating characteristic curve; LNH, lymph node harvest.
In summary, tumor size, in T1 colon cancer, showed a considerable predictive ability and outperformed any other widely used clinical factors, as compared with colon cancers of higher T stage.
Construction and Validation of Tumor Size–Based Nomogram in T1 Colon Cancer
To facilitate the clinical application of tumor size in survival prediction for patients with stage T1 colon cancer, we categorized tumor size according to its quartile and developed a nomogram in SEER data set based on multivariate Cox analysis. Tumor size (categorical), age, N stage, grade, and LNH were incorporated into this nomogram (Fig. 2A). Harrell's c‐index of this nomogram for CSS prediction in SEER was 0.710 (95% confidence interval, 0.628–0.791). Time‐dependent ROC suggested that the nomogram incorporating tumor size was significantly more accurate in predicting CSS than the model without tumor size (AUC at 5 years: 0.701 vs. 0.643; Fig. 2C).
Figure 2.
Construction and validation of tumor size–based nomogram in T1 colon cancer. (A): Tumor size–based nomogram in Surveillance, Epidemiology, and End Results (SEER) data set. (B): Nomogram without tumor size in SEER. (C): Time‐dependent ROC of the nomogram with or without tumor size. (D): Validation and prediction performance of the nomogram with or without tumor size in Fudan University Shanghai Cancer Center cohort.
Abbreviations: AUC, area under the ROC; LNH, lymph node harvest; ROC, receiver operating characteristic curve.
FUSCC cohort was further used to validate the predictive efficiency of the nomogram developed in the SEER database. These data suggested that a tumor size–based nomogram can predict CSS in FUSCC with a good discriminative ability, but absence of tumor size will notably decrease the prognostic accuracy (Fig. 2D).
Discussion
In this present study, the large population‐based SEER database and FUSCC database were used to analyze the impact of T stage on the prognostic and predictive value of tumor size in colon cancer. We demonstrated here that increase of T stage will significantly reduce the impact of tumor size on death and recurrence risk of colon cancer. In addition, tumor size outperformed any other clinical factors and showed the best discriminative ability in patients with stage T1 colon cancer but showed dramatically weakened discriminatory power in predicting CSS and DFS in patients with more advanced T stage.
Previous studies suggested that solid tumors including gastrointestinal tumors can obtain the potential of dissemination during the process of growing both horizontally and vertically 16. Currently, although many investigators have confirmed the negative prognostic role of tumor size, the value of tumor size in survival prediction is still underappreciated 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13. However, the vertical growth index reflected by T stage has been established as one of the dominating prognostic factors and has been incorporated into the widely accepted TNM staging system in colon cancer. Therefore, although both are a growth index, vertical infiltration seems to be much more important than horizontal proliferation, which might explain why the prognostic prediction value of tumor size dwindled as the T stage became more advanced. Like many previous studies 6, 8, 10, 17, we confirmed the independent predictive effect of tumor size in prognosis among all patients with stage I–III colon cancer. However, the HR of tumor size decreased gradually with the increase of T stage. In other words, more advanced tumor infiltration will weaken the risk impact on prognosis of larger tumor size. Similar results were found in the predictive power analysis. More advanced T stage is also able to negatively influence the survival predictive ability of tumor size. Once the tumor infiltrates into the muscular layer, the discriminative ability decreases dramatically.
One of the most important reasons for the negative impact of T stage on tumor size could be the inaccurate calculation of tumor size. It is imaginable that for the tumors with advanced stage, accurate measurement of tumor size is difficult because the invasion extent of the bowel wall might be larger than the maximal diameter of the cancerous extent on the mucosa. Therefore, merely the maximal horizontal diameter cannot exactly characterize the tumor growth extent with advanced infiltration. On the other hand, for the tumors restricted to the submucosa, growing into the lumen may be the main growth pattern, and the tumor size measured at this stage will be the dominant index that depicts the growth of tumor. Even so, we believe there are many other underlying causes for the impact on tumor size necessitating clarification.
An exciting result revealed in this study is that tumor size performed the best in survival prediction among many widely accepted clinical features in T1 stage colon cancer. T1 colon cancer is always characterized as an early stage tumor with favorable prognosis. Its long‐term survival depends mainly upon the lymph node status, with a colorectal cancer–specific 5‐year survival of ≥95% in the absence of metastatic lymph node and 68%–90% in the presence of positive lymph node 18, 19. However, the positive rate of lymph node was reported only in 8%–12% of patients 20, 21. In this present study, the lymph node metastatic rate was only 10.1% and 9.9% in patients with stage T1 colon cancer from SEER and FUSCC data sets, and less than 1% of patients were categorized as N2 stage. Consequently, it is reasonable to find that the discriminatory ability of N stage was weak in T1 colon cancer. Therefore, the predictive value of N stage should be reappraised according to T stage. Actually, in addition to N stage, many other predefined high‐risk clinical features, including poor differentiation, mucinous histology, and high CEA level, were infrequently found in T1 stage colon cancer 22, 23, 24, 25. Hence, using these factors alone to predict survival of T1 stage colon cancer may not be reliable.
To the best of our knowledge, the present study is one of the first study evaluating the impact of T stage on the prognostic and predictive value of tumor size in colon cancer. In addition, we revealed the critical role of tumor in predicting survival in T1 colon cancer. However, several limitations cannot be avoided completely in our study. To begin with, this study is a retrospective analysis, and selection bias could not be completely avoided. Furthermore, although this study was conducted in two large population databases, the proportion of T1 stage colon cancer is still too small in Chinese patients from the FUSCC database; thus, future multicenter studies are warranted to validate the results presented in this study.
Conclusion
The prognostic and predictive value of tumor size varied in patients with different T stages. The impact of tumor size on death and recurrence risk decreases gradually with the increment of T stage. More intriguingly, in patients with stage T1 colon cancer, tumor size possesses a better discriminative ability than any other established prognostic factors.
Author Contributions
Conception/design: Renjie Wang, Ye Xu, Guoxiang Cai
Provision of study material or patients: Qingguo Li, Ye Xu, Guoxiang Cai
Collection and/or assembly of data: Weixing Dai, Shaobo Mo, Wenqiang Xiang, Lingyu Han
Data analysis and interpretation: Weixing Dai, Shaobo Mo, Ye Xu, Guoxiang Cai
Manuscript writing: Weixing Dai, Shaobo Mo
Final approval of manuscript: Weixing Dai, Shaobo Mo, Wenqiang Xiang, Lingyu Han, Qingguo Li, Renjie Wang, Ye Xu, Guoxiang Cai
Disclosures
The authors indicated no financial relationships.
Acknowledgments
We appreciate the SEER program for making the database publicly available. This study was supported by the National Key R&D Program of China (2016YFC0905300 and 2016YFC0905301), the Grant of Science and Technology Commission of Shanghai Municipality (16401970502), the Grant of National Natural Science Foundation of China (81572351), Shanghai Shenkang Program (SHDC12014206), and the Development Fund for Shanghai Talents (2017120).
Disclosures of potential conflicts of interest may be found at the end of this article.
Contributor Information
Renjie Wang, Email: wangbladejay@sina.com.
Ye Xu, Email: yexufuscc@163.com.
Guoxiang Cai, Email: gxcaifuscc@163.com.
References
- 1. Brenner H, Kloor M, Pox CP. Colorectal cancer. Lancet 2014;383:1490–1502. [DOI] [PubMed] [Google Scholar]
- 2. Edge SB, Byrd DR, Compton CC et al. AJCC Cancer Staging Manual. 7th ed New York, NY: Springer, 2010. [Google Scholar]
- 3. Santullo F, Biondi A, Cananzi FCM et al. Tumor size as a prognostic factor in patients with stage IIa colon cancer. Am J Surg 2018;215:71–77. [DOI] [PubMed] [Google Scholar]
- 4. Spelt L, Sasor A, Ansari D et al. Pattern of tumour growth of the primary colon cancer predicts long‐term outcome after resection of liver metastases. Scand J Gastroenterol 2016;51:1233–1238. [DOI] [PubMed] [Google Scholar]
- 5. Steinberg SM, Barwick KW, Stablein DM. Importance of tumor pathology and morphology in patients with surgically resected colon cancer. Findings from the Gastrointestinal Tumor Study Group. Cancer 1986;58:1340–1345. [DOI] [PubMed] [Google Scholar]
- 6. Kornprat P, Pollheimer MJ, Lindtner RA et al. Value of tumor size as a prognostic variable in colorectal cancer: A critical reappraisal. Am J Clin Oncol 2011;34:43–49. [DOI] [PubMed] [Google Scholar]
- 7. Poritz LS, Sehgal R, Hartnett K et al. Tumor volume and percent positive lymph nodes as a predictor of 5‐year survival in colorectal cancer. Surgery 2011;150:649–655. [DOI] [PubMed] [Google Scholar]
- 8. Wang Y, Zhuo C, Shi D et al. Unfavorable effect of small tumor size on cause‐specific survival in stage IIa colon cancer, a SEER‐based study. Int J Colorectal Dis 2015;30:131–137. [DOI] [PubMed] [Google Scholar]
- 9. Balta AZ, Sucullu I, Saydam M et al. Horizontal tumor diameter as a prognostic factor. Am J Surg 2016;211:304–305. [DOI] [PubMed] [Google Scholar]
- 10. Saha S, Shaik M, Johnston G et al. Tumor size predicts long‐term survival in colon cancer: An analysis of the National Cancer Data Base. Am J Surg 2015;209:570–574. [DOI] [PubMed] [Google Scholar]
- 11. Moreno CC, Mittal PK, Sullivan PS et al. Colorectal cancer initial diagnosis: Screening colonoscopy, diagnostic colonoscopy, or emergent surgery, and tumor stage and size at initial presentation. Clin Colorectal Cancer 2016;15:67–73. [DOI] [PubMed] [Google Scholar]
- 12. Dai W, Li Y, Meng X et al. Does tumor size have its prognostic role in colorectal cancer? Re‐evaluating its value in colorectal adenocarcinoma with different macroscopic growth pattern. Int J Surgery 2017;45:105–112. [DOI] [PubMed] [Google Scholar]
- 13. Kato T, Alonso S, Muto Y et al. Tumor size is an independent risk predictor for metachronous colorectal cancer. Oncotarget 2016;7:17896–17904. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361–387. [DOI] [PubMed] [Google Scholar]
- 15. Heagerty PJ, Lumley T, Pepe MS. Time‐dependent ROC curves for censored survival data and a diagnostic marker. Biometrics 2000;56:337–344. [DOI] [PubMed] [Google Scholar]
- 16. Li F, Kishida T, Kobayashi M. Serum iron and ferritin levels in patients with colorectal cancer in relation to the size, site, and disease stage of cancer. J Gastroenterol 1999;34:195–199. [DOI] [PubMed] [Google Scholar]
- 17. Al Natour RH, Saund MS, Sanchez VM et al. Tumor size and depth predict rate of lymph node metastasis in colon carcinoids and can be used to select patients for endoscopic resection. J Gastrointest Surg 2012;16:595–602. [DOI] [PubMed] [Google Scholar]
- 18. Gunderson LL, Jessup JM, Sargent DJ et al. Revised TN categorization for colon cancer based on national survival outcomes data. J Clin Oncol 2010;28:264–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Chang GJ, Rodriguez‐Bigas MA, Skibber JM et al. Lymph node evaluation and survival after curative resection of colon cancer: Systematic review. J Natl Cancer Inst 2007;99:433–441. [DOI] [PubMed] [Google Scholar]
- 20. Ikematsu H, Yoda Y, Matsuda T et al. Long‐term outcomes after resection for submucosal invasive colorectal cancers. Gastroenterology 2013;144:551–559; quiz e514. [DOI] [PubMed] [Google Scholar]
- 21. Bosch SL, Teerenstra S, de Wilt JH et al. Predicting lymph node metastasis in pT1 colorectal cancer: A systematic review of risk factors providing rationale for therapy decisions. Endoscopy 2013;45:827–834. [DOI] [PubMed] [Google Scholar]
- 22. Asayama N, Oka S, Tanaka S et al. Long‐term outcomes after treatment for pedunculated‐type T1 colorectal carcinoma: A multicenter retrospective cohort study. J Gastroenterol 2016;51:702–710. [DOI] [PubMed] [Google Scholar]
- 23. Kobayashi H, Mochizuki H, Morita T et al. Characteristics of recurrence after curative resection for T1 colorectal cancer: Japanese multicenter study. J Gastroenterol 2011;46:203–211. [DOI] [PubMed] [Google Scholar]
- 24. Backes Y, Elias SG, Bhoelan BS et al. The prognostic value of lymph node yield in the earliest stage of colorectal cancer: A multicenter cohort study. BMC Med 2017;15:129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Suh JH, Han KS, Kim BC et al. Predictors for lymph node metastasis in T1 colorectal cancer. Endoscopy 2012;44:590–595. [DOI] [PubMed] [Google Scholar]