Adjuvant sandwich therapy is associated with improved survival in stage III–IVA endometrial cancer: a network meta-analysis

Abstract

Objective

Patients with advanced endometrial cancer (International Federation of Gynecology and Obstetrics stages III–IVA) face a high risk of recurrence even after surgery. Although adjuvant chemoradiotherapy (CRT) is standard care, the optimal sequencing strategy remains controversial. This research aimed to evaluate the impact of 4 CRT sequencing strategies on survival in patients with stages III–IVA endometrial cancer.

Methods

We conducted a network meta-analysis of 9 studies, encompassing 8,401 patients, to compare 4 CRT regimens: concurrent CRT (CCRT), chemotherapy followed by radiotherapy (CR), radiotherapy followed by chemotherapy (RC), and sandwich therapy involving chemotherapy-radiotherapy-chemotherapy (CRC). The primary outcomes were overall survival (OS) and progression-free survival (PFS), which were analyzed using risk ratios (RRs) with 95% confidence intervals (CIs). Treatment rankings were determined based on surface under the cumulative ranking (SUCRA) probabilities.

Results

CRC demonstrated superior survival outcomes compared to other regimens, with significantly higher mortality risks observed in comparator groups: for OS, CCRT showed a 99% increased risk (RR=1.99; 95% CI=1.28–3.09), CR 48% (RR=1.48; 95% CI=1.11–1.97), and RC 57% (RR=1.57; 95% CI=1.09–2.26); for PFS, CR exhibited 40% higher recurrence risk (RR=1.40; 95% CI=1.16–1.68) and RC 35% (RR=1.35; 95% CI=1.03–1.77). SUCRA values ranked CRC highest for both OS (92.98%) and PFS (96.32%), with no significant differences in grade ≥3 toxicity across regimens.

Conclusion

For patients with stages III–IVA endometrial cancer, CRC is associated with survival benefits without an increase in toxicity. Pending prospective validation, these findings support CRC as the preferred adjuvant regimen.

Synopsis

Sandwich therapy (chemotherapy-radiotherapy-chemotherapy, CRC) demonstrates superior overall and progression-free survival compared to other chemoradiotherapy modalities, without an increase in toxicity. Prospective randomized trials are recommended to validate CRC efficacy.

INTRODUCTION

Endometrial carcinoma (EC) is the most prevalent gynecological malignancy, comprising 37%–50% of all gynecological cancers, and its incidence is steadily increasing [1,2]. Registry data indicate that 67% of patients with early-stage localized disease amenable to surgical management can achieve favorable outcomes [3]. However, patients with advanced-stage EC (particularly stages III–IVA) experience recurrence rates of 45%–60% despite surgical interventions [4,5]. Patients with stage III–IV endometrial cancer are classified as a high-risk group, for whom combined adjuvant chemotherapy plus radiotherapy is recommended after surgery. The PORTEC-3 trial indicated that while patients with stage III EC treated with chemoradiotherapy (CRT) showed significantly superior 5-year progression-free survival (PFS) and decreased local recurrence rates compared to external beam radiation therapy (EBRT) alone [6]. For EC patients, the 2022 European Society for Medical Oncology (ESMO) guidelines recommend formulating postoperative adjuvant CRT strategies based on risk stratification to reduce tumor recurrence and improve survival rates [7].

However, there is still no definite conclusion on which CRT regimen is more beneficial for patients with stages III–IVA endometrial cancer [8]. Besides, finding from another study, GOG 258, showed no significant differences in 5-year overall survival (OS) or PFS between the CRT and chemotherapy-alone groups for patients with stages III–IVA EC, but CRT demonstrated better locoregional metastasis control [9]. Current reports indicate diverse CRT regimens, and the optimal sequencing of chemotherapy and radiotherapy remains controversial [10]. This uncertainty is partly due to the diversity in treatment goals, the heterogeneity in clinical trial designs (e.g., different tumor stages and pathological types), and, particularly, the lack of head-to-head randomized controlled trials (RCTs) comparing treatment approaches [11]. To address this clinical gap, this investigation employed network meta-analysis (NMA), representing the first comprehensive comparison of CRT sequencing efficacy. NMA integrates available evidence to enable simultaneous direct and indirect regimen comparisons [12]. Our research aimed to identify the sequence of radiotherapy and chemotherapy that could best balance their respective toxicities while reducing recurrence and improving survival.

MATERIALS AND METHODS

1. Study design and search strategy

This study conducted a systematic review and NMA, and is reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA)-NMA reporting guidelines [13]. We systematically searched PubMed and Embase, updating our search to include data up to July 1, 2025, and limiting it to studies published in English. The search terms included medical terminology such as: “concurrent chemoradiotherapy,” “sequential chemoradiotherapy,” “sandwich chemoradiotherapy,” “adjuvant chemoradiotherapy,” “chemoradiotherapy,” “endometrial cancer,” and “clinical trial.” The complete search strategy is presented in Data S1. Our search initially identified 256 articles (221 from PubMed and 35 from Embase), with 8 additional articles found through a review of the reference lists. Literature search flow diagram and detailed reasons for article exclusion are presented in Table 1. We used EndNote X9 and NoteExpress (Clarivate, Philadelphia, PA, USA) to manage the literature and remove duplicate records. Two researchers independently screened records, resolving any disagreements regarding inclusion through discussion.

Table 1. Main features of the eligible studies in the network meta-analysis.

Study	Study design	Country or area	Stage	Histology	Type of primary surgery	Adjuvant therapy	Adjuvant regimen	No. of patients	Follow-up (mo)	Median age (yr)
Secord et al. (2009) [20]	Retrospective	USA	III/IV	Endometrioid (48%); Pure papillary serous (17%); Clear cell (5%)	TH+BSO ± retroperitoneal lymphadenectomy	CRC	C: 6 TC or other (70%); 3 TC or other (12%); R: EBRT (36%); EBRT ± extended field and VBT (32%); VBT (5%)	45	33.6	61.5±11.2
						RC		18		61.3±9.2
						CR		46		62.0±10.5
Dogan et al. (2013) [27]	Retrospective	Turkey	IIIC	Endometrioid (72%); Pure papillary serous (20%); Clear cell (8%)	TH+BSO + retroperitoneal lymphadenectomy	CRC	C: 6 TC or other (72%); R: EBRT (72%)	11	18 (12–53)	63 (36–83)
						CR		14		61.5 (40–76)
Lu et al. (2015) [21]	Retrospective	USA	III	Endometrioid (100%)	TH+BSO ± retroperitoneal lymphadenectomy	CRC	C: ≥6 TC or other (76%); R: EBRT (100%)	14	55.2	56
						CR		37		57
Onal et al. (2018) [26]	Retrospective	Turkey	IIIC	Endometrioid (72%); Non-endometrioid (28%)	TH+BSO + retroperitoneal lymphadenectomy	CRC	C: 6 TC (80%); R: EBRT (100%) ± extended field and VBT	83	41 (5–167)	61 (36–88)
						CR		96
Latham et al. (2019) [25]	Retrospective	USA	IIIC	Endometrioid (60%); Clear cell (14.38%); Pure papillary serous (3.08%)	TH	CCRT	C: Unclear; R: EBRT (100%) ± extended field and VBT	1,169	-	43 (25–71)
						RC		5,116
						CR		696
McEachron et al. (2020) [24]	Retrospective	USA	III/IV	Endometrioid (44.0%); Serous (47.5%); Clear cell (8.5%)	TH+BSO + retroperitoneal lymphadenectomy	CRC	C: 4–8 TC (83.6%); R: EBRT ± extended field and VBT (60.5%); EBRT (39.5%)	56	66	66.0 (48–81)
						RC		52		64.9 (47–85)
						CR		44		65.6 (48–83)
Chen and Xiang (2020) [16]	Retrospective	China	III/IV	Endometrioid (77.6%); Pure papillary serous (19%); Clear cell (3.4%)	TH+BSO + retroperitoneal lymphadenectomy	CRC	C: AP (44%); TC (24.1%) or other ≥6 cycles (37%); R: EBRT + VBT (100%)	58	46.1±7.7	56.6±9.6
						CR		58	45.3±7.1	57.9±9.7
Narasimhulu et al. (2020) [23]	Retrospective	USA	IIIC	Endometrioid (52%); Non-endometrioid (48%)	TH+BSO ± retroperitoneal lymphadenectomy	CRC	C: ≥6 cycles (84.7%); R: EBRT ± extended field (42.9%); EBRT (55%)	18	70.8	63.8±10.6
						CR		84
Hathout et al. (2021) [22]	Retrospective	USA	IIIC	Endometrioid (66%); Non-endometrioid (34%)	TH+BSO ± retroperitoneal lymphadenectomy	CR	C: 6 TC (IQR, 5–6); R: EBRT+VBT (58%); EBRT (28%); VBT (14%)	369	45.3 (23.4–71)	62 (55–70)
						CCRT		113
						CRC		170
						RC		34

AP, cisplatin-adriamycin; CCRT, concurrent chemotherapy and radiation; CR, chemotherapy followed by radiotherapy; CRC, chemotherapy-radiotherapy-chemotherapy; EBRT, external beam radiation therapy; IQR, interquartile range; RC, radiotherapy followed by chemotherapy; TC, taxol+carboplatin; TH+BSO, total hysterectomy and bilateral salpingo-oophorectomy; VBT, vaginal brachytherapy.

2. Study selection

Trials were eligible if they met all the following criteria: (1) inclusion of patients with International Federation of Gynecology and Obstetrics (FIGO) 2009 stages III–IVA EC who underwent surgery; (2) administration of postoperative CRT; (3) comparison between 2 distinct CRT regimens; (4) reporting PFS and/or OS as outcome measures; (5) one of the following study designs: cohort studies, case-control studies, RCTs, or non-randomized clinical trials. Studies were excluded based on the following criteria: (1) duplicate publications, reviews, case reports, conference proceedings, animal/cell-based laboratory studies, single-arm trials, and inaccessible full texts; (2) incomplete outcome data; (3) implementation of concomitant therapies other than the predefined interventions.

3. Treatment groups

Our study categorized patients into 4 treatment groups based on the sequencing of adjuvant radiotherapy and chemotherapy: (1) the chemotherapy-radiotherapy-chemotherapy (CRC) group (sandwich therapy), which patients received 2–3 cycles of postoperative chemotherapy followed by radiotherapy (CR) and then an additional 3–4 cycles of chemotherapy; (2) the chemotherapy followed by radiotherapy (CR) group (sequential chemotherapy-first regimen), which patients completed 6 cycles of chemotherapy before radiotherapy; (3) the radiotherapy followed by chemotherapy (RC) group (sequential radiotherapy-first regimen), which radiotherapy was administered before commencing 6 cycles of subsequent chemotherapy; and (4) the concurrent chemoradiotherapy (CCRT) group (concurrent regimen), in which patients received 2 cycles of cisplatin concurrently with EBRT in the first and the last week, subsequently received 4 additional cycles of cisplatin and paclitaxel.

Based on the GOG 209 study results [14], carboplatin area under the curve 5 or 6 plus paclitaxel 175 mg/m² (TC) administered every 21 days for 4–6 cycles were recommended as first-line therapy for EC [15]. While most studies included in this analysis employed the TC regimen, one exception was the study by Chen and Xiang [16], where 44% of patients received the doxorubicin 40–60 mg/m² and cisplatin 70–75 mg/m² (AP) regimen. The RTOG 9708 phase II trial defined a CCRT regimen in which chemotherapy and radiotherapy were administered simultaneously. This regimen consisted of 2 cycles of cisplatin 50 mg/m² on weeks 1 and 4 of radiotherapy, followed by 4 cycles of cisplatin 50 mg/m² and paclitaxel 175 mg/m² administered at 21-day intervals [17].

Pelvic radiotherapy, typically 45–50.4 Gy, is administered 5 times weekly for 5 consecutive weeks. All included clinical trials used EBRT, with some patients also receiving vaginal brachytherapy (VBT) or additional radiation to surrounding fields based on their condition.

4. Data extraction

A standardized electronic data extraction form was developed, with specific criteria established through multidisciplinary team discussions and iterative preliminary testing. Two researchers independently extracted data from all included studies, with subsequent cross-verification. The extraction form captured: (1) study characteristics (first author’s name, publication year, study design, country); (2) patient demographics (FIGO stage, histologic type, sample size, age, follow-up duration); (3) intervention details (primary surgical approach, adjuvant therapy modalities, chemoradiation agents, dosages, and treatment cycles); and (4) outcome measures.

The primary endpoints were PFS and OS, with effect sizes expressed as risk ratios (RRs). Event counts for these endpoints were defined as the number of patients experiencing death or disease progression, respectively. Secondary endpoints included treatment-related adverse events and recurrence patterns (overall, locoregional, distant). For adverse events, effect sizes were calculated as RR, with event counts based on patients manifesting these events. Recurrence patterns were quantified by recurrence rates, with corresponding event counts based on patients experiencing relapse.

5. Quality assessment

The Newcastle-Ottawa Scale (NOS) was used to assess non-randomized study quality (maximum score: 9). Studies scoring ≥7, 5–6, and <5 were categorized as high, medium, and low quality, respectively. Only those scoring ≥5 were included in the meta-analysis.

6. Statistical analysis

Data analysis was performed using R (version 4.4.3; R Foundation for Statistical Computing, Vienna, Austria) and the netmeta package [18], implementing graph-theoretical NMA within a frequentist paradigm. This paradigm primarily employs weighted least squares to compare relative effects among treatments. Two-sided statistical significance was set at p<0.05.

Random-effects model was used when significant heterogeneity (I²≥50%) was found, and fixed-effects model was used when non-significant heterogeneity (I²<50%) was detected. We assessed inconsistencies among interventions forming closed loops using a dual approach: (1) Global inconsistencies were evaluated using the Global Q-statistic based on generalized linear models, with p<0.05 indicating significant network-level inconsistency [19]; (2) Local inconsistencies were examined using the node-splitting method, where p<0.05 signified significant disagreement between direct and indirect evidence for specific treatment comparisons.

Treatment effects in survival analyses were quantified using RRs with corresponding 95% confidence intervals (CIs). The CRC group served as the common reference comparator in OS, PFS, and adverse event comparisons. An RR <1 indicated superior survival benefits over CRC, whereas RR >1 suggested the compared therapy represented a risk factor relative to CRC therapy. Statistical non-significance was defined as a 95% CI crossing 1.

We used the surface under the cumulative ranking (SUCRA) metric to rank interventions by their probability of being optimal (Rank 1), second-best (Rank 2), or least effective (Rank 4). SUCRA converts ranking probabilities into a continuous 0–1 scale, quantifying each treatment’s relative performance; values approaching 1 indicate superior overall ranking, while those near 0 reflect poorer performance. These relationships were visualized using cumulative ranking probability bar charts, where taller bars indicated higher probabilities of achieving a superior SUCRA rank position. Pearson’s χ² test was used to determine differences in the rates of overall, local, and distant recurrence.

Comparison-adjusted funnel plots for the primary outcomes (OS and PFS) were generated using the netmeta package. Publication bias was assessed by integrating formal visual symmetry evaluation with Egger’s linear regression test, where statistical significance (p<0.05) indicated potential bias.

RESULTS

1. Study selection

Preliminary screening identified 264 relevant articles. After removing duplicates and articles deemed irrelevant to the current study, we retained 45 publications for full-text screening. Ultimately, 9 studies, encompassing 8,401 patients, were included in the analysis [16,20,21,22,23,24,25,26,27]. Fig. 1 illustrates the study selection process.

Fig. 1. Flowchart of identified, included, and excluded studies from the network meta-analysis.

The included studies focused on the 4 commonly used sequencing methods investigated in this study: CCRT, CRC, CR, and RC. Comparisons between CRC and CR were the most frequent (n=5), followed by 3-way comparisons of CRC, CR, and RC (n=2). One study compared CCRT, CR, and RC, and another involved a comprehensive 4-way comparison of CCRT, CRC, CR, and RC. All 9 studies had a retrospective design, and most included patients were 43–66 years old. All patients underwent total hysterectomy with or without bilateral adnexectomy and pelvic lymphadenectomy, followed by postoperative CRT. The radiotherapy regimens predominantly involved EBRT ± extended-field irradiation or VBT, with doses ranging from 45 to 60 Gy. Chemotherapy protocols typically consisted of 6 cycles of carboplatin and paclitaxel. Table 1 summarizes the detailed characteristics of the included studies.

The NOS classified 4 studies as high-quality and 5 as moderate-quality (Table S1).

2. Network plots

The network plot (Fig. 2) compares the studies included in this analysis. The network exhibited sparsity, with no direct comparisons between CCRT and CRC in OS.

Fig. 2. Network plots. This figure presents (A) overall survival and (B) progression-free survival. Each node represents a treatment. All interventions are interconnected, and no isolated nodes present. The size of the nodes corresponds to the sample size of the respective therapy. Directly compared therapies are connected by black lines, where line thickness is proportional to the number of trials.

CCRT, concurrent chemoradiotherapy; CR, chemotherapy followed by radiotherapy; CRC, chemotherapy-radiotherapy-chemotherapy; RC, radiotherapy followed by chemotherapy.

3. OS and PFS

Using CRC as the reference in the overall population, the CCRT group exhibited significantly elevated mortality risk for OS (RR=1.99; 95% CI=1.28–3.09; p=0.021), representing a 99% increase. Similarly, both CR (RR=1.48; 95% CI=1.11–1.97; p=0.0069) and RC (RR=1.57; 95% CI=1.09–2.26; p=0.016) groups demonstrated significantly higher mortality risks—corresponding to 48.1% and 57.0% increases respectively—with both regimens demonstrating inferior survival outcomes relative to CRC.

Regarding PFS, the CR cohort demonstrated a 40% elevation in recurrence risk (RR=1.40; 95% CI=1.16–1.68; p=0.0004), and the RC group exhibited a 35% significant increase (RR=1.35; 95% CI=1.03–1.77; p=0.029; Fig. 3). The CCRT regimen was excluded from this analysis due to insufficient event data. Among the comparable regimens, CRC maintained clinical superiority with the lowest aggregate progression rate (17.2%), followed by RC (25.3%) and CR (26.9%).

Fig. 3. Forest plots presenting the network RR estimates, using sandwich therapy (CRC) as the comparison intermediary group. This figure displays (A) overall survival and (B) progression-free survival.

CCRT, concurrent chemoradiotherapy; CI, confidence interval; CR, chemotherapy followed by radiotherapy; CRC, chemotherapy-radiotherapy-chemotherapy; RC, radiotherapy followed by chemotherapy; RR, relative risk.

OS (I²=36.1%) and PFS (I²=0%) outcomes demonstrated non-significant heterogeneity, both falling below the conventional 50% threshold.

The ranking table (Fig. S1) presents a comparative analysis of treatment modalities (CRC, CCRT, CR, RC) across clinical outcomes (OS and PFS). CRC led to significantly higher OS than the other therapies (RR <1, CI excludes 1), achieving 1.47-fold higher efficacy than CR (1/0.68) and 1.56-fold higher efficacy than RC (1/0.64), while CR and RC showed statistically comparable efficacy. CCRT ranked the lowest in overall efficacy, but was statistically comparable to CR and RC. For PFS, CRC demonstrated a 39% higher efficacy over CR (1/0.72) and a 35% higher efficacy than RC (1/0.74), whereas CR and RC were statistically comparable.

As shown in Fig. 4, CRC had the highest probability of ranking first in OS (92.98%), followed by CR (63.38%) and RC (51.78%). CCRT showed the highest probability of ranking last (70.74%). CRC also had the highest probability of ranking first in PFS (96.32%), followed by RC (49.34%) and CR (53.02%).

Fig. 4. Accumulative bar charts of rank probabilities. This figure shows (A) overall survival and (B) progression-free survival. The bar charts illustrate the surface under the cumulative ranking, with the X-axis representing treatment regimens and the Y-axis indicating the probability of each treatment to achieve a specific rank (ranging from 0 to 1).

CCRT, concurrent chemoradiotherapy; CR, chemotherapy followed by radiotherapy; CRC, chemotherapy-radiotherapy-chemotherapy; RC, radiotherapy followed by chemotherapy.

Figs. 4 and S1 demonstrate that CRC yielded statistically superior efficacy in reducing both mortality risk and recurrence probability. Regarding OS, the ranking from best to worst was CRC, CR, RC, and CCRT, while it was CRC, RC, and CR for PFS.

Global heterogeneity tests resulted in p>0.05 for all, detecting no significant heterogeneity or inconsistency. The network demonstrated global consistency, indicating high credibility of the results. The node-splitting method detected no significant inconsistency (p>0.05), indicating the reliability of the network model (Fig. S2).

4. Recurrence site

Supplement 5 (S5) compares recurrence site outcomes among the CRT regimens. Differences in the frequencies of overall, locoregional, and distant recurrences were identified using Pearson’s χ² test. The studies by McEachron et al. [24] and Hathout et al. [22] reported only overall recurrence rates without specifying recurrence sites. Other articles provided detailed recurrence locations, which we categorized as local and distant recurrences for separate comparisons. In Table S2, statistically significant differences were observed in overall (χ²=22.91; p<0.001) and local (χ²=19.95; p<0.001) recurrence rates, but not in distant metastasis recurrence rates (χ²=2.06; p=0.357). RC had the highest overall (50.96%) and local (44.44%) recurrence rates.

5. Adverse reaction

Fig. 5 shows a comparison of the adverse reactions associated with the CRT regimens. Common treatment-related toxicities of anticancer drugs in this study primarily included gastrointestinal, genitourinary, and hematological (anemia, neutropenia, and thrombocytopenia) toxicity.

Fig. 5. Forest plot presenting the RR of adverse reactions compared with sandwich therapy (chemotherapy-radiotherapy-chemotherapy).

CI, confidence interval; CR, chemotherapy followed by radiotherapy; GUS, Genitourinary System; GIS, Gastrointestinal System; RC, radiotherapy followed by chemotherapy; RR, relative risk.

The heterogeneity tests for adverse reaction rates across studies were not statistically significant (I²<50%) for CRC, RC, and CR; therefore, a fixed-effects model was applied. The forest plots demonstrate that CR, RC, and CCRT treatment modalities had statistically comparable adverse reaction rates to CRC.

6. Risk of bias

Publication bias was assessed by examining the symmetry of the funnel plot (Fig. S3). The plot appeared largely symmetrical, and Egger’s test showed no significant evidence of publication bias (p>0.05), suggesting no substantial publication bias influenced the NMA results.

DISCUSSION

This NMA compared the efficacy of 4 commonly used CRT regimens using data from 9 retrospective studies. With CRC as the reference comparator, our study revealed that CRC was significantly superior to the other regimens, ranking the highest in both OS and PFS while maintaining a low recurrence risk and not increasing the adverse event rate.

A previous meta-analysis by Zhang et al. [28] demonstrated significantly superior 5-year OS with the CRC regimen compared to sequential CRT (corresponding to CR and RC in this study; hazard ratio=1.33; 95% CI=1.09–1.60; p=0.004). However, our work extends this evidence by integrating CCRT, following the 2022 ESMO guidelines, into the comparison. No head-to-head efficacy comparisons between CCRT and CRC in adjuvant treatment have been reported. Therefore, our evaluation is the first to contrast CCRT, CR, RC, and CRC, providing critical insights for optimizing postoperative management in EC.

Our research detected a significant advantage for CRC in OS and PFS. This finding highlights the superior efficacy of CRC in both survival outcomes, whereas CCRT showed the worst performance in OS. Given its superior survival benefits, we advocate postoperative CRC as the preferred clinical option for patients with stages III–IVA EC. The CR and RC regimens showed statistically comparable OS, PFS, and adverse event profiles. However, the recurrence rate following CR was lower than following RC, indicating a potential relative advantage of CR over RC in locoregional disease control. We found no significant differences regarding the tolerance of the different radiotherapy and chemotherapy sequences. These findings emphasize the consistent OS and PFS advantages of CRC and the inferiority of CCRT.

According to the American Society for Radiation Oncology clinical practice guidelines, all patients with stages III–IVA EC have indications for postoperative radiotherapy. Adjuvant radiotherapy after surgery helps eradicate micrometastatic foci and optimizes local control, significantly reducing the risks of vaginal and/or pelvic recurrence [29]. Maggi et al. [30] demonstrated that postoperative chemotherapy delayed distant metastases and markedly reduced para-aortic lymph node recurrences. Radiotherapy appeared more effective in controlling locoregional disease, whereas chemotherapy showed superior control over distant dissemination [31].

The therapeutic advantage of CRC over CCRT lies in its ability to ensure completion of the full radiation and chemotherapy dosage while avoiding CCRT’s characteristic acute toxicity amplification [32]. The RC sequence delays chemotherapy initiation relative to CRC, creating a postoperative window for microscopic residual disease to progress. Conversely, the CR approach risks compromising radiotherapy feasibility due to cumulative toxicity from 6 consecutive chemotherapy cycles that often deplete the patient’s functional reserves before radiotherapy can be delivered. CRC strategically overcomes these limitations through phased chemotherapy delivery: administering initial post-surgery chemotherapy cycles for timely micrometastasis control, followed by radiotherapy during a chemotherapy-free interval to minimize overlapping toxicities, and concluding with consolidation chemotherapy. This sequenced approach simultaneously reduces cumulative toxicity burdens, maintains treatment intensity, preserves radiotherapy tolerance, and ultimately achieves dual suppression of locoregional recurrence and distant metastasis through optimized therapeutic coordination. This phased sequencing ensures full treatment intensity while avoiding both CCRT’s acute toxicity and RC/CR’s progression or tolerance risks. Our heterogeneity and inconsistency tests indicated no significant bias in transmission. The robustness of NMA estimates further substantiates our findings.

While our results suggest potential survival disadvantages of CCRT compared to CRC (RR=1.99; 95% CI=0.51–242 0.94; p=0.021), this NMA had some limitations. First, our study included no RCTs and clinical trials that directly compared the regimens. All included studies were retrospective, and indirect comparisons were generated through NMA. Studies comparing CCRT to other CRT regimens were lacking, and critical data on treatment-related adverse events and recurrence rates were unavailable in the original literature. Second, subgroup analyses based on pathological and molecular subtypes were not feasible due to limited data granularity. Consequently, the efficacy of the 4 regimens across distinct histological subtypes remains unexplored.

Our findings suggest that CRC is a more tolerable and prognostically favorable sequencing strategy than other CRT regimens. Head-to-head RCTs and multicenter, prospective studies are needed to validate our findings and enhance the reliability and generalizability of our results. With sufficient data, further studies could explore precision treatments guided by pathological and molecular subtyping.

This NMA suggests CRC as a potentially optimal postoperative regimen. However, this finding requires validation in future RCTs and multi-center clinical studies.

SUPPLEMENTARY MATERIALS

Data S1

Search strategy

Table S1

The Newcastle-Ottawa Quality Assessment Scale of the included non-randomized controlled studies

Table S2

Recurrence rates of different interventions

Fig. S1

Network league table. This figure presents (A) overall survival and (B) progression-free survival. Each cell indicates the relative risk and 95% confidence interval for row-based versus column-based therapies. The bold values indicate statistically significant differences.

Fig. S2

Node-split forest plot for inconsistency assessment. (A) Overall survival and (B) progression-free survival.

Fig. S3

Funnel plot.