Survival outcomes in IIIC cervical cancer by treatment strategies: a systematic review and meta-analysis

Abstract

Objective

To compare survival outcomes of patients with stage IIIC cervical cancer based on different treatment strategies.

Method

Studies reporting survival of cervical cancer patients in IIIC stage were identified through systematic searches of the PubMed, Web of Science, CNKI and Wanfang data, and the hazard ratios (HRs) of survival outcomes were calculated based on different treatment strategies: radical hysterectomy (RH) group versus radical chemoradiotherapy (RCRT) group. Heterogeneity in the pooled data was assessed using the I² statistic.

Results

We meta-analyzed data from nine studies involving 6,793 patients, who were assigned to different treatment strategies: RH group, 3840 patients (56.5%) and RCRT group, 2,953 patients (43.5%). High heterogeneity was observed across studies. Compared to the RCRT group, the RH group showed similar overall survival (HR 0.77, 95% CI 0.58–1.04, I² =83.4%), disease-free survival (HR 0.68, 95% CI 0.23–2.04, I² =88.3%) and progression-free survival (HR 0.67, 95% CI 0.10–4.36, I² =91.4%).

Conclusions

Current evidence suggests no statistically significant survival difference between RCRT and RH in IIIC cervical cancer. These findings don’t contradict the NCCN recommendation, but the high heterogeneity and wide confidence intervals warrant cautious interpretation due to potential variations in patient populations or treatment protocols.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14697-6.

Introduction

The incidence of cervical cancer has declined significantly in recent years, owing to the implementation of effective screening programs, HPV vaccination, and treatment of precancerous lesions. Despite these advancements, cervical cancer remains a major global health burden, with approximately 600,000 new cases and 340,000 deaths reported in 2020 [1, 2]. While prognosis is favorable in early-stage cervical cancer, it deteriorates when the disease invades surrounding tissues or metastasizes to regional lymph nodes. Lymph node metastasis (LNM) is one of the most important prognostic factors [3, 4], with studies showing a 30–50% reduction in 5-year survival rates and increased recurrence risk [5, 6].

In 2018, the International Federation of Gynecology and Obstetrics (FIGO) revised its staging system to incorporate lymph node metastasis status, classifying pelvic and para-aortic lymph node involvement as stage IIIC1 and IIIC2 [7], respectively. However, recent studies have revealed inconsistencies between this updated staging system and the prognosis of stage IIIC disease. Specifically, stage IIIC1 patients with isolated lymph node involvement demonstrate better survival rates compared to those with stage IIIA or IIIB disease, highlighting prognostic heterogeneity among stage IIIC patients [8–11].

Treatment strategies for cervical cancer patients have evolved over time. According to the 2009 FIGO staging system, radical hysterectomy (RH) with pelvic lymphadenectomy was the primary treatment for early-stage disease, including cases with LNM [3, 12]. Surgery offers precise staging and potential locoregional control advantages, particularly for patients with ≤ 2 positive lymph nodes [13]. The 2024 National Comprehensive Cancer Network (NCCN) guidelines, however, recommend radical chemoradiotherapy (RCRT) with adjuvant radiotherapy as standard care, given its dual efficacy in controlling both primary and metastatic lesions [14]. However, these recommendations are mainly based on studies published between 1999 and 2000 [15–19], which may not fully reflect contemporary treatment strategies. Furthermore, the guidelines do not address the heterogeneity of prognosis among stage IIIC patients, leaving a gap in personalized treatment strategies. Studies report 5-year overall survival (OS) rates of 50% − 60% for RCRT and 40% -75% for surgery [8, 20]. Current evidence mostly comes from single-center retrospective studies, which limits direct comparison.

This meta-analysis aims to compare survival outcomes among stage IIIC cervical cancer patients, based on different treatment strategies (RH group versus RCRT group). By evaluating the efficacy of these treatment strategies, this study seeks to provide evidence-based insights to guide clinical decisions for this heterogeneous patient population.

Materials and methods

The protocol for this systematic review and meta-analysis was registered in PROSPERO (CRD42025621533), and procedures were carried out according to the “Preferred Reporting Items for Systematic Reviews and Meta-analyses” (PRISMA) checklist [21].

Literature search and study inclusion

We systematically searched PubMed, Web of Science, CNKI and Wanfang data to identify relevant studies published from database inception through November 30, 2024. The search terms were tailored as follows: (‘lymph node metastasis’ OR ‘IIIC’) AND (‘cervical cancer’ OR ‘cervix neoplasm’ OR ‘Uterine Cervical Neoplasms’) AND (‘treatment’ OR ‘prognosis’ OR ‘outcome’ OR ‘surgery’ OR ‘radiotherapy’ OR ‘neoadjuvant chemotherapy’ OR ‘chemoradiotherapy’). Potential studies were also identified by manually searching the reference lists of included articles.

Eligible for inclusion were cohort studies, case-control studies or randomized controlled trials in which cervical cancer patients in stage IIIC according to the FIGO 2018 staging system were divided into the radical hysterectomy with pelvic lymphadenectomy + postoperative chemoradiotherapy group (RH group) and the radical chemoradiotherapy ± adjuvant therapy group (RCRT group). Studies had to compare overall survival (OS), disease-free survival (DFS) and/or progression-free survival (PFS) based on different treatment strategies in their samples. If studies examined overlapping patient samples, only the study scoring highest in the quality assessment was included.

Studies were excluded if they were case reports, unreviewed dissertations, meetings or reviews; did not report necessary data; scored < 7 stars in the quality assessment (see below); were not published in English or Chinese.

Assessment of study quality

Two researchers (Yi-xiang Li and Si-yu Cao) independently evaluated the methodological quality of included studies using the Newcastle-Ottawa Scale (NOS) [22]. For the item “Comparability of cohorts on the basis of the design or analysis”, studies that controlled for primary lesion size of cervical cancer were assigned one star, while studies that also controlled for the count of positive lymph nodes were assigned two stars. For the item “Adequacy of follow up of cohorts”, studies for which > 90.0% of patients were followed up received one star. Any disagreements were resolved through discussion with the corresponding author. Only studies that received at least seven stars altogether were included in the meta-analysis.

Data extraction

Two researchers (Yi-xiang Li and Si-yu Cao) independently extracted data from the included studies, including: first author, year of publication, sample size, clinicopathological features and survival data. OS was defined as the time from diagnosis to all-cause death; PFS as the time from diagnosis to progression; and DFS as the time from diagnosis to recurrence or metastasis. Any discrepancies were resolved through consensus discussion with the corresponding author.

Statistical analysis

Data were analyzed using Stata version 12.0 (StataCorp, College Station, TX, USA). For studies that reported OS, DFS or PFS only as Kaplan-Meier curves, we extracted survival data using Engauge Digitizer version 4.1 (http://sourceforge.net/projects/digitizer/) [23]. Survival analysis was performed using either fixed-effects (I² < 50%) or random-effects (I² >50%) model to calculate pooled hazard ratios (HR) with 95% confidence intervals (CIs). Heterogeneity was assessed using the I² statistic. When I² exceeded 50%, potential sources of heterogeneity were explored through Galbraith plot analysis [24]. Result robustness was assessed using leave-one-out sensitivity analyses, sequentially excluding each study to repeat the meta-analyses.

Begg’s test was used to assess publication bias [25]. All tests were two-tailed, with p-values < 0.05 considered statistically significant.

Results

From the initial 19,538 publications identified, we removed 10,941 duplicates, leaving 8,597 records for title and abstract screening. After full-text review of 27 articles, nine were included in the final meta-analysis [26–34], encompassing 6,793 patients (Fig. 1). The RH group comprised 3,840 patients (56.5%), and the RCRT group included 2,953 (43.5%). Sample sizes across the nine studies ranged from 30 to 4,086 patients, with geographic distribution as follows: China (four studies), USA (three), the Czech Republic (one), and the Netherlands (one).

Fig. 1.

Flow diagram of literature searching and study inclusion

The average age was 48 years in the RH group and 54 years in the RCRT group. Stratification by the AJCC 9th edition TNM staging system indicated a predominance of T1 (45.7%, n = 2,866) and T2 (37.1%, n = 2,332) disease, while T3 cases accounted for 17.2% (n = 1,081). Histologically, squamous cell carcinoma accounted for 88.5% (n = 6,004) of cases, whereas adenocarcinomas (8.2%) and adenosquamous carcinomas (3.3%) were less frequent. Among 283 cases, positive lymph node counts were reported, 57.5% had ≥ 2 positive lymph nodes. Among evaluable tumors (n = 1,407), 932 (57.9%) were ≤ 4 cm, and 465 (28.6%) were > 4 cm; the remaining 13.5% had unreported sizes. Additional baseline characteristics are summarized in Table 1.

Table 1.

Characteristics of patients in the included studies

	Cibula (2021)	Derks (2017)	Feng (2023)	Li (2023)	Potter (1990)	Richard (2008)	Ye (2023)	Zhang (2021)	Zhong (2024)	Total
Country	Czech Republic	Netherlands	China	USA	USA	USA	China	China	China
RH group	361 (70.1)	89 (73.6)	324 (66.4)	540 (50.0)	15 (50.0)	163 (74.8)	2173 (53.2)	27 (64.3)	148 (69.5)	3840 (56.5)
RCRT group	154 (29.9)	32 (26.4)	164 (33.6)	540 (50.0)	15 (50.0)	55 (25.2)	1913 (47.8)	15 (35.7)	65 (30.5)	2953 (43.5)
Total no.patients	515	121	488	1080	30	218	4086	42	213	6793
Age (Mean years)a
RH group	48	44	56	NR	NR	41	48	41	NR	48
RCRT group	46	44	58	NR	NR	36	55	43	NR	54
Histology
SCC	391 (76.5)	89 (73.5)	488 (100.0)	851 (78.8)	28 (93.3)	175 (81.8)	3761 (92.0)	34 (82.9)	187 (87.8)	6004 (88.5)
AC	87 (17.0)	26 (21.5)	0 (0.0)	167 (15.5)	2 (6.7)	21 (9.8)	230 (5.7)	5 (12.2)	20 (9.4)	558 (8.2)
ASC	33 (6.5)	6 (5.0)	0 (0.0)	62 (5.7)	0 (0.0)	18 (8.4)	95 (2.3)	2 (4.9)	6 (2.8)	222 (3.3)
T stageb
T1	381 (77.9)	109 (90.1)	NR	817 (75.6)	27 (90.0)	218 (100.0)	1117 (27.3)	35 (83.3)	162 (76.1)	2866 (45.7)
T2	108 (22.1)	12 (9.9)	NR	263 (24.4)	3 (10.0)	0 (0.0)	1888 (46.2)	7 (16.7)	51 (23.9)	2332 (37.1)
T3	0 (0.0)	0 (0.0)	NR	0 (0.0)	0 (0.0)	0 (0.0)	1081 (26.5)	0 (0.0)	0 (0.0)	1081 (17.2)
Positive lymph nodes
< 2	NR	NR	NR	NR	8 (26.7)	NR	NR	13 (31.0)	98 (46.0)	119 (41.8)
≥ 2	NR	NR	NR	NR	20 (66.6)	NR	NR	29 (69.0)	115 (54.0)	164 (57.5)
Missing	NR	NR	NR	NR	2 (6.7)	NR	NR	0 (0.0)	0 (0.0)	2 (0.7)
Lesion size
≤ 4 cm	251 (48.7)	93 (76.9)	210 (43.0)	NR	24 (80.0)	116 (53.2)	NR	35 (83.3)	213 (100.0)	942 (57.9)
> 4 cm	97 (18.8)	28 (23.1)	278 (57.0)	NR	5 (16.7)	50 (22.9)	NR	7 (16.7)	0 (0.0)	465 (28.6)
Missing	167 (32.5)	0 (0.0)	0 (0.0)	NR	1 (3.3)	52 (23.9)	NR	0 (0.0)	0 (0.0)	220 (13.5)

Values are n, n (%). AC, adenocarcinoma; ASC, adenosquamous carcinoma; SCC, squamous cell carcinoma; NR, not reported; RH, radical hysterectomy; RCRT, radical chemoradiotherapy. ^aThe numbers represent the mean ages of the radical hysterectomy (RH) and radical chemoradiotherapy (RCRT) groups. ^bAccording to TNM staging system in the 9th version of the American Joint Committee on Cancer (AJCC) staging manual for cervical cancer, patients were divided into three groups: T1, T2 and T3

The follow-up duration across studies ranged from 45 to 96 months. All included studies were retrospective cohort studies, with most demonstrating low risk of bias in survival outcomes according to the NOS criteria. Five studies [27–29, 31, 32] exhibited bias due to confounding, while four studies [26, 28, 30, 33, 34] showed bias in outcome assessmen. Additionally, one study demonstrated bias due to inadequate cohort follow-up (Fig. 2) [33].

Fig. 2.

Risk of bias in included studies based on the NOS criteria in Review Manager software

In the meta-analysis, nine studies reported HRs for OS. The pooled HR for OS was 0.77 (95% CI 0.58–1.04, p = 0.093, I² = 83.4%), using the RCRT group as the reference (Fig. 3). Subgroup analyses by publication year, age, region, sample size, T-stage distribution, and lesion size did not identify significant sources of heterogeneity (Supplementary Fig S1). Galbraith plot (Supplementary Fig S2A) identified two studies [29, 32] as potential outliers. Sensitivity analysis excluding these studies confirmed the original results (HR 0.81, 95% CI 0.63–1.04, p = 0.105, I² = 43.6%) (Supplementary Fig S2B). Notably, the studies by Li et al. (n = 1,080) and Ye et al. (n = 4,086) enrolled larger cohorts than the other included studies, with Ye et al. additionally analyzing T3-stage patients. Three studies calculated HRs for DFS involving 4,420 patients. The pooled HR for DFS was 0.68 (95% CI 0.23–2.01, p = 0.486, I² = 83.4%) [27, 32, 34] (Fig. 4). Only two studies calculated HRs for PFS involving 530 patients. The pooled HR for PFS was 0.67 (95% CI 0.10–4.36, p = 0.672, I² = 91.4%) [28, 33] (Fig. 5). We did not investigate the sources of heterogeneity in the pooled HRs for DFS and PFS, as few studies provided HRs for these endpoints.

Fig. 3.

Forest plot of pooled hazard ratios (HRs) for overall survival (OS) comparing treatment strategies: radical hysterectomy (RH) versus radical chemoradiotherapy (RCRT)

Fig. 4.

Forest plot of pooled hazard ratios (HRs) for disease-free survival (DFS) comparing treatment strategies: radical hysterectomy (RH) versus radical chemoradiotherapy (RCRT)

Fig. 5.

Forest plot of pooled hazard ratios (HR) for progression-free survival (PFS) comparing treatment strategies: radical hysterectomy (RH) versus radical chemoradiotherapy (RCRT)

We assessed publication bias only in the meta-analysis of OS because data came from all nine studies. Begg’s test indicated no obvious bias (p = 0.917) (Supplementary Fig S3). Sensitivity analysis suggested that the HRs calculated for OS were robust (Supplementary Fig S4).

Discussion

Our meta-analysis found no statistically significant survival differences between RH and RCRT (OS HR 0.77, 95% CI: 0.58–1.04; DFS HR 0.68, 95% CI: 0.23–2.04; and PFS HR 0.67, 95% CI: 0.10–4.36), with substantial heterogeneity (I² >50%). These findings don’t contradict the NCCN recommendation of chemoradiotherapy as standard management for stage IIIC disease. However, the management of stage IIIC disease remains a controversy [16]. Prior to the 2018 FIGO staging revision, surgery was standard care for early-stage cervical cancer with lymph node involvement. Some studies advocate for RH as it can lead to better locoregional control and lower morbidity. Khalifa et al.‘s retrospective analysis of 145 cervical cancer cases revealed that the surgery group had significantly lower complications (28.3% vs. 61.7%, P < 0.001) compared to chemoradiotherapy group [35, 36]. Surgery can also achieve negative surgical margins and provide detailed pathological features from surgical specimens [37], which is important for tailoring adjuvant therapies [38]. For young women, RH is especially recommended due to its potential to preserve ovarian function and sexual function. Khalifa et al.‘s results showed 0.0% (surgery) versus 25.0% (chemoradiotherapy) vaginal stenosis incidence (P < 0.001) [36, 39, 40]. However, RH is associated with perioperative complications, including hemorrhage, infection, urinary tract injuries and long-term complications such as bladder dysfunction and lymphedema. Surgery is not suitable for patients with extensive involvement, distant metastases or poor performance status. Additionally, achieving complete resection in cases of bulky or fixed lesions can be challenging. Moreover, some patients who undergo surgery may require adjuvant chemoradiotherapy due to high-risk pathological features, potentially exposing them to cumulative toxicities from both treatments.

The NCCN recommendation reflects composite considerations including: (i) superior locoregional control in EMBRACE trial data (3-year locoregional control 89.0% vs. 76.0%) [41], (ii) avoidance of cumulative surgical/radiation morbidity, and (iii) better standardization of radiation techniques compared to surgical variability across institutions. Large randomized trials, including those conducted by the Gynecologic Oncology Group (GOG) [15], have demonstrated that RCRT achieves similar or better survival outcomes than RH in locally advanced cervical cancer. However, RCRT carries significant toxicity risks (15.0–25.0% grade ≥ 3 bowel/vaginal complications) that impair quality of life [41, 42]. Treatment response heterogeneity, which is influenced by tumor volume, hypoxia, and histology, may require salvage interventions in 20.0–30.0% of cases [43]. Unlike surgery, RCRT lacks pathological staging data, limiting precision in prognostic stratification.

To the best of our knowledge, this may be the first systematic review and meta-analysis of a relatively large patient population in this field, filling the evidence gap on the effects of different treatment strategies for stage IIIC disease. This meta-analysis tested the rationality of the 2024 NCCN recommendation for RCRT as the standard management for stage IIIC disease.

Our findings should be interpreted with caution in light of several limitations. One is that the included studies were retrospective studies, which may lead to potential biases. The included studies did not uniformly report detailed adjuvant therapies. Heterogeneity may arise from differences in chemoradiotherapy protocols, including the dose, duration and drug selection. Studies from high-income countries typically employed advanced radiotherapy techniques (IGABT/IMRT) and strict protocol compliance, while studies from resource-limited regions often used simpler techniques (2D brachytherapy) [41]. Treatment allocation (RH vs. RCRT) was influenced by institutional preferences (78.0% of RH cases in high-volume centers vs. 22.0% in low-volume centers, p < 0.001) and temporal trends (32.0% decline in surgery rates post-FIGO 2018) [44, 45].

Second, our meta-analysis did not compare differences in the treatment-related toxicities and functional outcomes. The choice between RH and RCRT for stage IIIC cervical cancer is complex, hinging not only on survival but also on toxicity burden and organic function that impact quality of life (QoL). Previous studies indicate that RCRT is associated with higher acute toxicity (35.0% vs. 18.0% with RH), including hematologic/gastrointestinal effects [44]. Chronic complications of RCRT also impair QoL: vaginal stenosis (61.5%, 25.0% grade ≥ 3), radiation proctitis (15.0–20.0%) [46], and dyspareunia (72.0% at 2 years) [44]. Conversely, RH preserves vaginal elasticity but causes lymphedema (28.3% vs. 3.1%, P < 0.001) and urinary incontinence (21.7% vs. 9.2%, P = 0.019) [47]. Since our meta-analysis did not comprehensively evaluate prognosis, future studies should integrate both survival equivalence and patient-centered outcomes to optimize the long-term QoL.

Third, only one study included in the meta-analysis explored the ability of neoadjuvant chemotherapy (NACT) followed by radical hysterectomy to improve prognosis of stage IIIC disease. NACT reduces tumor volume, thereby increasing the resection rates of tumor lesions and reducing the risk of postoperative recurrence, which may reduce lymph node lesions and improve locoregional control. Additionally, emerging therapies such as immune checkpoint inhibitors [48] and targeted agents (e.g., bevacizumab) [49], though promising in advanced or recurrent cervical cancer, were not included in our analysis. Future research should explore the role of immunotherapy and targeted therapy in stage IIIC disease to fill current evidence gaps.

Fourthly, inconsistent reporting of substage (IIIC1 vs. IIIC2) and primary tumor size (T1-T3) across studies may obscure clinically relevant differences in treatment response. The EMBRACE-II subgroup analysis showing worse 3-year PFS for IIIC2 vs. IIIC1 (62.0% vs. 74.0%, P = 0.03) (NCT03617133). A cohort study involved 9,452 cases revealed that patients with IIIC-T1/T2 (tumor ≤ 4 cm) exhibited a significantly higher 5-year survival rate compared to those with IIIC-T3 (tumor > 4 cm or pelvic wall invasion) (P < 0.05) [50]. Additionally, cases with T3 tumors have 40.0% lower pCR rates to neoadjuvant therapy than T1 (P < 0.01) [51].

Finally, the included studies did not provide the status of para-aortic lymph nodes, which is an important prognostic factor. The 2024 NCCN guidelines emphasize the importance of para-aortic lymphadenectomy (PAL) for accurate staging and treatment planning in stage IIIC disease. Data from GOG trials (GOG85, GOG120, GOG165) [15, 52, 53] suggest that patients with pathologically confirmed Stage IIIC2p disease have a better prognosis than those with radiologically defined Stage IIIC2r disease, highlighting the potential prognostic benefit of PAL. Further studies are needed to clarify its therapeutic impact.

Collectivelly, these limitations reveal critical knowledge gaps in balancing efficacy with patient-centered outcomes. Our findings underscore the imperative for prospective studies implementing: standardized IIIC substage classification, comprehensive toxicity profiling, QoL evaluation, and rigorous assessment of modern therapies (NACT, targeted agents, and immunotherapy). Addressing these priorities will optimize stage IIIC cervical cancer management amidst evolving treatment paradigms.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Contribution of authorship is as follows: Conceptualization: Yi-xiang Li, Jin-ke Li. Data curation: Yi-xiang Li, Si-yu Cao. Formal analysis: Yi-xiang Li, Yu-Fan. Investigation: Yi-xiang Li. Methodology: Yi-xiang Li. Project administration: Jin-ke Li. Supervision: Jin-ke Li. Writing - original draft: Yi-xiang Li, Si-yu Cao, Yu-fei Zhang. Writing - review & editing: Jin-ke Li, Si-yu Cao.

Funding

Not applicable.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.