Real-world analysis of survival benefit of surgery and adjuvant therapy in elderly patients with colorectal cancer

Treatment guidelines for colorectal cancer (CRC) in elderly patients remain unclear. This study aimed to investigate whether elderly patients (≥ 70 years) with CRC benefit from surgery and adjuvant therapy. A total of 90,347 eligible CRC patients older than 70 years were collected from the Surveillance, Epidemiology, and End Results (SEER) database and divided into a surgery group and a no-surgery group. After being matched by propensity score matching at a 1:1 ratio, 23,930 patients were included in our analysis. The Kaplan‒Meier method and log-rank test were applied to compare overall survival (OS) and cancer-specific survival (CSS). Univariate and multivariate Cox regression analyses were utilized to confirm independent prognostic factors for OS and CSS. In age-stratified analysis (70–74; 75–79; 80–84; ≥ 85), the OS and CSS rates of patients in the surgery group were significantly higher than those of patients in the no-surgery group (all P < 0.001). Adjuvant therapy was an independent prognostic factor for OS and CSS in elderly patients with CRC (all P < 0.001). Further analysis showed that elderly colon cancer patients with stage III and stage IV disease gained a survival benefit from adjuvant chemotherapy. Adjuvant chemoradiotherapy can significantly improve OS and CSS in elderly rectal cancer patients with stage II, III, and IV disease. In conclusion, among CRC patients aged ≥ 70 years reported in the SEER database, treatment with surgical resection is significantly associated with improved OS and CSS. Moreover, adjuvant therapy led to a significant prognostic advantage for elderly advanced CRC patients who underwent surgery.


Methods
www.nature.com/scientificreports/All eligible patients were divided according to whether patients received surgery into a surgery group and a no-surgery group.No surgery was defined as no surgery of the primary site, only autopsy, or only excisional biopsy.Our study included all patients who met the inclusion and exclusion criteria to minimize selection bias.Propensity score analysis was used to control for potential confounding by the matching factors.Patients were propensity matched 1:1 into surgery and no-surgery groups through the nearest neighbour method with a calliper of 0.1 times the standard deviation of the propensity score.Variables used for matching were as follows: age, sex, year of diagnosis, race, marital status, tumour site, histological type, differentiated grade, radiotherapy (RT), and CT.

Statistical analysis.
All analyses were performed with IBM SPSS Statistics 25.0 software (IBM, Armonk, NY, USA) and R version 4.0.0 (http:// www.r-proje ct.org), and two-tailed P values < 0.05 were assessed as statistically significant.To balance covariance and reduce the bias of efficacy evaluation, we performed 1:1 PSM between the surgery group and no surgery group.Standardized differences were used to examine the balance across baseline covariates before and after matching, and a standardized difference below 10% was considered to be sufficiently reliable to provide well-balanced covariates after matching.Next, chi-square analysis was utilized to compare clinicopathologic characteristics in both matched and unmatched groups.A Cox regression model was applied for multivariate analysis to identify significant prognosticators.We calculated odds ratios (ORs) and 95% confidence intervals (CIs) for each prognosticator.The Kaplan-Meier method and log-rank test were used to compare overall survival (OS) and cancer-specific survival (CSS).

Results
Baseline characteristics.After screening, a total of 90,347 patients were included in the final analysis.
Among them, 78,120 (86.47%) patients were enrolled in the surgery group and 12,227 (13.53%) in the nosurgery group.After ten characteristics were matched by PSM at a 1:1 ratio, 23,930 patients were included in our analysis.The baseline characteristics of the patients before and after PSM are summarized in Supplemental Table 1.We found significant differences between the groups regarding age, sex, year of diagnosis, race, marital status, tumour site, histological type, differentiated grade, RT, and CT among the clinicopathologic characteristics of the surgery and no-surgery groups before PSM.After matching by the propensity score, the baseline characteristics were similar for the matched cohort for most variables, except for age and CT.Estimates of standardized difference scores after matching were all less than 10%, which indicates that the baseline patient characteristics were balanced after matching (Supplementary Fig. 1).

Survival outcomes of surgical patients and nonsurgical patients after PSM.
To compare the survival rate of the surgery and no-surgery groups at different ages, we divided the 23,930 CRC patients into four age stages (ages 70-74; 75-79; 80-84; ≥ 85): 5455, 5390, 5576, and 7509 patients aged 70-74, 75-79, 80-84, and ≥ 85, respectively.In CRC patients treated with surgery, the 5-year OS rates for those aged 70-74, 75-79, 80-84, and ≥ 85 years were 63.1%, 55.3%, 47.7%, and 35.2%, respectively, which were clearly better than those of patients not treated with surgery (Fig. 2A-D).Similarly, in CRC patients treated with surgery, the 5-year CSS rates for those aged 70-74, 75-79, 80-84, and ≥ 85 years were 72.6%, 70.9%, 66.4%, and 60.9%, respectively, significantly better prognoses than those of patients not treated with surgery (Fig. 2E-H).OS and CSS rates in CRC patients treated with surgery in each age group were significantly higher than those in the no-surgery group (all P < 0.001).Subgroup analysis according to tumour location was also performed.Overall, surgery, whether for colon or rectal cancer, can significantly improve prognosis (Supplementary Figs. 2, 3).www.nature.com/scientificreports/Univariate and multivariate Cox analyses in patients after PSM.After PSM, multivariate Cox analysis showed that surgery (OR = 3.601, 95% CI 3.485-3.721,P < 0.001) was an independent predictive factor for OS (Supplemental Table 2).Analogously, the results of multivariate Cox analysis indicated that surgery (OR = 4.688, 95% CI 4.497-4.886,P < 0.001) correlated significantly with CSS (Supplemental Table 3).Remarkably, the OR of surgery was highest, and surgical treatment was the most significant protective factor for both OS and CSS in older patients with CRC.We also performed subgroup analysis by tumour location (Supplemental Tables 4-7).Similar results were also found for colon and rectal cancer analysed separately.TNM staging was not included in multivariate analysis because an accurate pathological stage could not be obtained for nonsurgical patients.Thus, we further compared the survival outcomes of surgical patients in different stages with nonsurgical patients.Compared with stage IV patients who underwent surgery, nonsurgical patients had a worse outcome (Supplementary Figs. 4, 5).
Univariate and multivariate Cox analyses for patients who underwent surgery after PSM.Further univariate and multivariate analyses were performed for patients who underwent surgery after PSM.Due to the different prognoses and therapeutic patterns of colon and rectal cancer, we performed subgroup analysis based on tumour site.In patients with colon cancer, multivariate Cox analysis showed that CT (OR = 1.187, 95% CI 1.082-1.301,P < 0.001) was an independent predictive factor for OS (Supplemental Table 8).However, CT was not significantly associated with the CSS of colon cancer patients in multivariate Cox analysis (Supplemental Table 9).

Survival outcomes of different treatment patterns of patients who underwent surgery at different stages after PSM.
We conducted subgroup analysis stratified by stage to further analyse the influence of different treatment patterns on prognosis.In stage I colon cancer patients, the 5-year OS rates were 55.4% and 62.7% for patients with surgery + CT and surgery alone, respectively (P = 0.585, Fig. 3A); the 5-year CSS rates were 73.3% and 87.3% for patients with surgery + CT and surgery alone, respectively (P = 0.003, Fig. 3E).In patients with stage II disease, the 5-year OS rates were 58.8% and 48.1% for patients with surgery + CT and surgery alone, respectively (P = 0.051, Fig. 3B), and the 5-year CSS rates were 70.3% and 75.6% for patients with surgery + CT and surgery alone, respectively (P = 0.017, Fig. 3F).In patients treated with surgery + CT, the 5-year OS rates for patients with stage III and stage IV disease were 54.7% and 13.2%, respectively, which were significantly better prognoses than those of patients treated with surgery alone (all P < 0.001, Fig. 3C,D).Similarly, in patients treated with surgery + CT, the 5-year CSS rates for patients with stage III and stage IV disease were 66.7% and 14.8%, respectively, also significantly better prognoses than those of patients treated with surgery alone (all P < 0.001, Fig. 3G,H).
In rectal cancer patients with stage I disease, the 5-year OS rates were 65.8% and 55.2% for patients with surgery + chemoradiotherapy (CRT) and surgery alone, respectively (P = 0.013, Fig. 4A), and the 5-year CSS rates were 81.0% and 82.0% for patients with surgery + CRT and surgery alone, respectively (P = 0.563, Fig. 5A).In patients treated with surgery + CRT, the 5-year OS rates for patients with stage II, stage III, and stage IV disease were 64.5%, 54.5% and 22.1%, respectively, which were significantly better prognoses than those of patients treated with surgery alone (all P < 0.001, Fig. 4B-D).Among patients treated with surgery + CRT, the 5-year   CSS rates of stage II, stage III, and stage IV were 75.8%, 63.2%, and 26.8%, respectively, with significantly better prognosis than surgery alone (all P < 0.001, Fig. 5B-D).
In rectal cancer patients with stage I disease, the 5-year OS rates were 46.4% and 55.2% for patients with surgery + CT and surgery alone, respectively (P = 0.149, Fig. 4E); the 5-year CSS rates were 72.2% and 82.0% for patients with surgery + CT and surgery alone, respectively (P = 0.112, Fig. 5E).In patients with stage II disease, the 5-year OS rates were 60.3% and 38.1% for patients with surgery + CT and surgery alone, respectively (P < 0.001, Fig. 4F); the 5-year CSS rates were 63.3% and 61.6% for patients with surgery + CT and surgery alone, respectively (P = 0.788, Fig. 5F).In patients treated with surgery + CT, the 5-year OS rates for patients with stage III and stage IV disease were 51.5% and 16.3%, respectively, significantly better prognoses than those of patients treated with surgery alone (all P < 0.001, Fig. 4G,H).In patients treated with surgery + CT, the 5-year CSS rates for patients with stage III and stage IV disease were 60.3% and 18.2%, respectively, significantly better prognoses than those of patients treated with surgery alone (all P < 0.001, Fig. 5G,H).
In patients treated with surgery + RT, the 5-year OS rates for patients with stage I, II, III, and IV disease were 46.4%, 41.5%, 26.9% and 30.3%, respectively; these results indicated no significant improvement in prognosis compared with patients treated with surgery alone (all P > 0.05, Fig. 4I-L).Similarly, the 5-year CSS rates for patients treated with surgery + RT with stage I, II, III, and IV disease were 77.2%, 58.3%, 39.4% and 20.0%, respectively, which did not show a significant improvement in prognosis compared with patients treated with surgery alone (all P > 0.05, Fig. 5I-L).

Survival outcomes of different treatment patterns of patients who underwent surgery in different age groups after PSM.
Age-stratified analysis was also conducted to analyse the influence of different treatment patterns on prognosis.In stage III and stage IV colon cancer patients treated with surgery + CT, the 5-year OS rates for patients aged 70-74, 75-79, 80-84, and ≥ 85 years were 50.9%, 49.0%, 41.5%, and 28.7%, respectively; these rates were clearly better than those of patients treated with surgery alone (Fig. 6A-D).Similarly, in patients treated with surgery + CT, the 5-year CSS rates for patients aged 70-74, 75-79, 80-84, and ≥ 85 years were 55.9%, 55.7%, 50.6%, and 39.2%, respectively, which were significantly better prognoses than those of patients treated with surgery alone (Fig. 6E-H).

Discussion
The incidence of CRC is known to increase with age.As life expectancy is increasing, the number of elderly CRC patients with decreased functional status and increased comorbidity is also expected to increase worldwide 2 .However, the relatively high risks of complications, postoperative sequelae, in-hospital mortality, and sensitivity to treatment toxicity in elderly patients with CRC have led to refusal of surgery and aggressive adjuvant treatments.In the present study, we carefully identified appropriate patients from the 17 SEER databases and investigated therapeutic patterns and survival outcomes.Our results showed that surgical resection was associated with significantly longer OS and CSS in elderly CRC patients regardless of how old the patients were.Furthermore, we found that surgery + CT was associated with an apparent survival benefit in elderly advanced colon cancer Surgery is widely believed to be the best approach for treating solid tumours 16 .However, decisions regarding whether surgery should be performed for elderly patients are sometimes difficult for clinicians, patients, and their families, as influenced by numerous factors.A retrospective multicentre study in Japan demonstrated that age was a factor that reduced the survival of patients with resected CRC (P < 0.001) 17 .Kocián et al. reported that compared with young patients, the poorer prognosis of elderly patients (≥ 75 years) is presumably due to their worse physiological reserve and higher incidence of comorbidities 18 .However, one systematic review concluded that the CSS of rectal cancer patients does not decrease with age 19 .Thus, age alone should not be a decisive factor in medical decisions.Additional information is likely needed for patients to decide whether to undergo surgery.To our knowledge, studies of whether elderly CRC patients should undergo surgical therapy have not yet been fully explored.In our research, PSM was conducted to balance baseline characteristics related to survival outcome.After matching, surgical therapy was an independent prognostic factor in multivariate Cox regression analysis, and further survival analysis showed that patients in the surgery group had better OS and CSS than those in the no-surgery group in different age groups.Therefore, surgical treatment may be prudently recommended for elderly CRC patients.Careful clinical observation and assessment of elderly patients, acceptable morbidity, anticipated life span, comorbidities, nutrition, and functional status should be considered when deciding surgical cure.
Adjuvant therapy to target occult metastasis is advocated for stage II cancer with high-risk features and stage III disease 20 .However, the standard adjuvant treatment regimens for elderly patients are currently unclear because patients aged ≥ 70 years are frequently excluded from randomized trials 21 .Moreover, some studies have reported that the clinical effectiveness of adjuvant CT in elderly CRC patients appears to be limited, with increased chemotherapy-related toxic side effects 22,23 .Thus, it is necessary to analyse the efficacy of adjuvant therapy in elderly CRC patients, which is a special group of cancer patients.Sargent et al. combined data from seven randomized controlled trials (RCTs) and found that adjuvant CT had a significant positive effect on both OS and time to tumour recurrence for colon cancer and that age did not appear to affect this result 24 .In general, RCTs have strict eligibility inclusion and exclusion criteria and typically exclude elderly patients with comorbidities and high frailty 25 .Nevertheless, real-world studies, with less restrictive study populations, might be more reflective of the effectiveness of interventions in real settings.A systematic review of real-world studies suggested that adjuvant CT significantly improves OS but not disease-free survival (DFS) in elderly CRC patients 26 .Our research found that adjuvant CT was significantly associated with improved OS and CSS in stage III and stage IV colon cancer patients.This result converges with research derived from younger patients 27,28 .However, our study was not able to demonstrate a prognostic benefit of adjuvant CT in elderly stage II colon cancer patients.This may be attributable to the inability to differentiate stage II patients into high-and low-risk groups.Many studies have demonstrated that risk factor stratification may contribute to efficient selection of stage II colon cancer patients who will benefit from adjuvant therapy 29,30 .Our study also found that stage I patients who received CT had worse prognosis than those who did not.Stage I CRC patients have a very low risk of recurrence, and thus postoperative adjuvant therapy is not indicated.This phenomenon may be due to postoperative tumour recurrence or metastasis.
Neoadjuvant chemoradiotherapy (nCRT) can downstage locally advanced rectal cancer, which results in a lower rate of postoperative local recurrence and is now considered standard treatment for locally advanced rectal cancer worldwide 31 .Adjuvant CRT is currently offered routinely after radical resection for patients with rectal cancer receiving nCRT, though the role of adjuvant therapy in rectal cancer remains controversial.EORTC 22921 evaluated the effect of neoadjuvant and/or adjuvant CT on preoperative RT and radical resection 32 , with no significant impact of adjuvant CT on OS, DFS or the cumulative incidence of distant spread for the whole group.The Italian I-CNR-RT trial randomized 655 patients with locally advanced rectal cancer to 6 cycles of 5FU vs. observation after nCRT and surgery 33 .They also concluded no benefit for adjuvant CT in the 5-year OS, distant metastasis, and local recurrence.To our knowledge, there are no RCTs or other high-quality studies investigating the efficacy of nCRT and adjuvant CT in rectal cancer patients over 70 years.In our research, CRT was significantly associated with improved OS and CSS in stage II, stage III, and stage IV rectal cancer patients in all age groups.However, limited data prevented us from ascertaining whether CRT was administered before or after surgery in rectal cancer patients.Thus, future clinical trials should further explore the value of nCRT and adjuvant CRT in older adults with advanced rectal cancer.
According to recent data, the life expectancy of medically fit elderly patients is gradually increasing 34 .Therefore, a new diagnosis of older adult CRC patient has the potential to increase life expectancy if treated.Regardless, the physiological heterogeneity of elderly patients with frequent discrepancies between biological and chronological age is coupled with additional complications.The International Society of Geriatric Oncology recommends that elderly CRC patients undergo preoperative evaluation of biological age based on the physiological side effects of ageing, physical and mental ability, and social support 35 .Thus, it is crucial for clinicians to distinguish fit from more vulnerable older patients.Moreover, outcomes related to surgery and adjuvant therapy should be well considered, such as postoperative morbidity and mortality, functional status, quality of life, toxicity of therapy, completion of therapy, and survival.A comprehensive geriatric assessment system should be established to aid in such therapeutic decisions.After adequate evaluation, treatment can be adjusted to maximize effectiveness and better meet the individual requirements of older patients.As a systematic review of seven existing frailty screening methods in older patients receiving surgery, RT, or CT concluded insufficient discriminative power to improve patient selection 36 , further research is required to establish a more efficient geriatric biological age evaluation system in oncology that can evaluate benefit/risk ratios of various treatment interventions for patients 37 .
Although the SEER database provides public data to investigate such clinical problems, there are several limitations in this study.First, treatment information, including RT dosing, CT regimen, and the sequence of CT, was not recorded in the SEER database, and we did not analyse the role of these factors.In addition, SEER records did not include information about locoregional relapse or distant metastasis; thus, we were unable to assess local recurrence survival and DFS.Further research directions include appropriate CT regimens used to treat elderly CRC patients, the effect of postoperative adjuvant therapy on elderly rectal cancer patients who receive neoadjuvant therapy, and screening appropriate elderly CRC patients who will benefit from adjuvant therapy.

Conclusion
The current study indicates that surgical resection is significantly associated with improved prognosis for CRC patients aged ≥ 70 years in 17 SEER databases.Moreover, adjuvant therapy showed a significant prognostic advantage for elderly advanced CRC patients who underwent surgery in all age groups.We look forward to future high-quality prospective clinical studies that can reliably validate our conclusions.

Figure 1 .
Figure 1.Flowchart for data filtration of older patients with colorectal cancer according to SEER dataset.PSM, propensity score matching.

Figure 3 .
Figure 3. Kaplan-Meier survival analysis of colon cancer patients underwent surgery after propensity score matching (surgery + chemotherapy vs. surgery alone).Overall survival for patients with I stage (A), II stage (B), III stage (C), and IV stage (D).Cancer-specific survival for patients with I stage (E), II stage (F), III stage (G), and IV stage (H).CT, chemotherapy.

Figure 4 .
Figure 4. Survival analysis for overall survival of rectal cancer patients underwent surgery after propensity score matching.Overall survival for patients with I stage (A), II stage (B), III stage (C), and IV stage (D) (surgery + CRT vs. surgery alone).Overall survival for patients with I stage (E), II stage (F), III stage (G), and IV stage (H) (surgery + CT vs. surgery alone).Overall survival for patients with I stage (I), II stage (J), III stage (K), and IV stage (L) (surgery + RT vs. surgery alone).CRT, chemoradiotherapy; CT, chemotherapy; RT, radiotherapy.

Figure 5 .
Figure 5. Survival analysis for cancer-specific survival of rectal cancer patients underwent surgery after propensity score matching.Cancer-specific survival for patients with I stage (A), II stage (B), III stage (C), and IV stage (D) (surgery + CRT vs. surgery alone).Cancer-specific survival for patients with I stage (E), II stage (F), III stage (G), and IV stage (H) (surgery + CT vs. surgery alone).Cancer-specific survival for patients with I stage (I), II stage (J), III stage (K), and IV stage (L) (surgery + RT vs. surgery alone).CRT, chemoradiotherapy; CT, chemotherapy; RT, radiotherapy.