Prognostic value and adjuvant treatment outcomes for the use of the JGOG prognostic model in early-stage intermediate-risk cervical cancer: a Korean Gynecologic Oncology Group study (KGOG 1028)

Abstract

Objective

The aim of this study was to validate a new Japanese risk grouping in the Korean population and to identify the group benefit from adjuvant treatment.

Methods

A total of 561 patients with stage IB cervical cancer who underwent radical hysterectomy with lymphadenectomy from 2000 to 2008 across 9 Korean Gynecologic Oncology Group-affiliated institutions were included in this study. Patients had at least one intermediate-risk factor: lymphovascular space invasion, outer one-third of the deep cervical stromal invasion, or a tumor 4 cm or larger. Disease-free survival (DFS) was analyzed based on adjuvant therapies: no treatment, radiation therapy, concurrent chemoradiation therapy or systemic chemotherapy.

Results

Patients were classified into 3 groups based on their histologically-incorporated intermediate risk factors. Group 1 included patients with limited risk factors (n=385, 68.6%), group 2 (n=106, 18.9%), and group 3 (n=70, 12.5%) had increasing risk. DFS differed across groups (group 1 vs. group 2, p=0.048; group 2 vs. group 3, p=0.030). Group 1 showed no DFS benefit from adjuvant treatment, while in groups 2 and 3, receiving radiation therapy improved DFS (group 2: adjusted hazard ratio [aHR]=0.09; 95% confidence interval [CI]=0.01–0.80; p=0.031 and group 3: aHR=0.21; 95% CI=0.04–1.08; p=0.047). In group 3, receiving concurrent chemotherapy treatment did not significantly affect DFS compared to radiation alone (aHR=0.84; 95% CI=0.11–6.30; p=0.861).

Conclusion

This study revealed the prognostic value of histology-incorporated intermediate-risk stratification for early-stage cervical cancer in the Korean population. Unlike the results seen in Japanese populations, adjuvant treatment benefited higher-risk patients in terms of DFS.

Synopsis

This study demonstrated that the Japanese Gynecologic Oncology Group prognostic model for early-stage intermediate-risk cervical cancer has prognostic value in a Korean population. Adjuvant treatment benefited higher-risk patients, suggesting that this risk scoring system may help guide adjuvant treatment decisions.

INTRODUCTION

Cervical cancer is the fourth most common cancer in women worldwide, with approximately 660,000 new cases and approximately 350,000 deaths in 2022 [1]. With the increased early screening for cervical cancer through the use of Pap smears and human papillomavirus tests, the number of patients diagnosed with cervical cancer at early stages is also on the rise [2]. Although surgery is a primary treatment for early-stage cervical cancer, the recurrence rate after radical hysterectomy may be as high as 30% [3,4].

Adjuvant radiation and concurrent chemoradiation therapy have been suggested to improve disease-free survival (DFS) after surgery. Traditionally, the surgical and pathological intermediate risk factors for cervical cancer include invasion of the lymphovascular space, deep cervical stromal invasion, and a large tumor size [5]. The Gynecologic Oncology Group (GOG) 92 study demonstrated that, for stage IB cervical cancer patients with 2 or more intermediate risk factors, receiving radiation therapy after radical hysterectomy improved DFS compared with no further treatment [3]. A 2024 Japanese GOG (JGOG) study introduced a classification-tree model with a 3-tier risk scoring system, which sequentially arranges the risk levels for each risk factor alongside the patient’s histology [6,7]. The results of this study revealed that, in terms of recurrence-related oncologic outcomes, the omission of adjuvant treatment after radical hysterectomy was not significantly worse compared to receiving adjuvant treatment for all patients included in this study.

The JGOG prognostic model has not yet been validated in cohorts outside of Japan. Furthermore, considering the discrepancies in perspectives on postoperative adjuvant treatment between the GOG and JGOG studies, further validation is necessary. Thus, the objective of this study was to develop and validate a new Japanese risk scoring system in the Korean population that incorporates histology and intermediate-risk factors. Second, we sought to identify the risk group that would benefit from adjuvant treatment.

MATERIALS AND METHODS

1. Patients

From January 2000 to December 2008, a total of 2,205 patients with early-stage cervical cancer from 9 Korean Gynecologic Oncology Group (KGOG)-affiliated institutions were reviewed retrospectively after this study received approval from each institutional review board. Patients were pathologically confirmed to have a diagnosis of International Federation of Gynecology and Obstetrics stage IA–IIA cervical cancer, and received radical hysterectomy (type II or III) with pelvic and/or paraaortic lymphadenectomy.

Patients with stage IB cervical cancer were reviewed in this study. The study cohort was restricted to patients with any of the following pathologically confirmed invasion of the lymphovascular space, invasion of the outer one-third of the deep cervical stroma, or the presence of a tumor 4 cm or larger. Patients who were treated with neoadjuvant chemotherapy or radiation therapy before surgery, who were incidentally confirmed to have cervical cancer after simple hysterectomy, or who had insufficient data or rare histology were excluded. Patients with pathologically advanced disease, such as parametrial invasion, pelvic or paraaortic lymph node metastasis, or pelvic or distant metastases, were also excluded. As a result, from a database, 561 patients satisfied the inclusion criteria (Fig. 1).

Fig. 1. Flowchart of included patients and definition of histology-incorporated intermediate-risk group.

^*Pathologically confirmed parametrial invasion, lymph node metastasis or resection margin positive.

^†Lympho-vascular space invasion, outer one-third deep cervical stromal invasion, or tumor size of 4 cm or larger.

2. Exposure and outcome

The adjuvant therapy modalities after radical hysterectomy with lymphadenectomy were as follows: no treatment, radiation therapy, concurrent chemoradiation therapy and systemic chemotherapy alone. The choice of treatment modality varied depending on the preferences of each institution and physician. This study also used 5-year DFS as the outcome measure, following the approach employed in the JGOG study. DFS was defined as the time interval from surgery to the first evidence of any recurrence.

3. Analysis

First, we followed the histology-incorporated intermediate risk model for recurrence suggested in the previous GOG 92 and JGOG studies and classified all enrolled patients into 3 groups. Histological findings were divided into 2 group; squamous cell carcinoma (SCC) or others. The surgical-pathological intermediate-risk groups were classified as follows: invasion of the lymphovascular space (present or absent), deep cervical stromal invasion (outer one-third, inner or middle one-third), and tumor size (larger than 4 cm, equal to or smaller than 4 cm). DFS within each group was analyzed on the basis of postoperative adjuvant treatment, excluding patients who received chemotherapy only. DFS curves were constructed via the Kaplan–Meier method and compared with the log-rank test (equal weights at all event times). The association with postoperative therapy was analyzed via a Cox proportional hazard regression model estimating the adjusted hazard ratio (aHR) and the corresponding 95% confidence interval (CI). The multivariate Cox model was adjusted for age and tumor size. Statistical analyses were performed using R 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria). A p-value of less than 0.05 was considered significant.

RESULTS

1. Patient characteristics

The median age of the patients was 43 years, and most patients had SCCs (n=425, 75.8%). The median tumor size was 3 cm. A total of 320 patients (57.0%) were had deep cervical stromal invasion, and a total of 318 patients (56.7%) had no lymphovascular invasion. Approximately half of the patients (n=235, 41.9%) received no postoperative adjuvant treatment, followed by 178 patients (31.7%) who received radiation therapy, 132 patients (23.5%) who received concurrent chemoradiation therapy, and 16 patients (2.9%) who received only chemotherapy (Table 1).

Table 1. Demographic and pathologic data of the patients.

Characteristics		Total (n=561)
Age (yr)		43 (23–78)
FIGO stage
	IB1	441 (78.6)
	IB2	120 (21.4)
Histology
	SCC	425 (75.8)
	Non-SCC	136 (24.2)
Tumor size (cm)		3 (0–8.5)
Lymphovascular space invasion
	Negative	243 (43.3)
	Positive	318 (56.7)
Invasion depth
	Inner 1/3	104 (18.5)
	Middle 1/3	137 (24.4)
	Deep 1/3	320 (57.0)
Postoperative therapy modality
	Postoperative surveillance	235 (41.9)
	Radiation therapy	178 (31.7)
	Concurrent chemoradiation therapy	132 (23.5)
	Chemotherapy only	16 (2.9)

Values are presented as median (range) or number (%).

FIGO, International Federation of Gynecology and Obstetrics; SCC, squamous cell carcinoma.

2. Patient classification and DFS for each group

For the histologically-confirmed intermediate risk factors, group 1 included patients with no histological evidence of invasion of the lymphovascular space or histological evidence of SCC with invasion of the lymphovascular space and inner or middle cervical stromal invasion (n=385, 68.6%). Group 2 included patients with evidence of SCC and invasion of the lymphovascular space with deep cervical stromal invasion (n=106, 18.9%). Group 3 included patients with nonsquamous histology and invasion of the lymphovascular space (n=70, 12.5%) DFS curves were plotted for these groups (p<0.001) (Fig. 2).

Fig. 2. Disease-free survival stratified by histology-incorporated intermediate-risk groups. (A) Patient classification based on histology-incorporated intermediate-risk and associated DFS. (B) Kaplan-Meier curves comparing DFS across histology-incorporated intermediate-risk groups.

DFS, disease-free survival; DSI, deep cervical stromal invasion; LVSI, lympho-vascular space invasion; SCC, squamous cell carcinoma.

3. Survival outcome and its association with DFS based on postoperative therapy

In each group, patients who received chemotherapy were excluded, and DFS was analyzed on the basis of postoperative adjuvant treatment. The p-values for each group are as follows: 0.28 for group 1, 0.03 for group 2, and 0.021 for group 3 (Fig. 3). Additionally, in group 2, compared with patients without postoperative treatment, patients who received radiation therapy had a significantly lower aHR (0.09; 95% CI=0.01–0.80; p=0.031) . In Group 3, both radiation therapy and concurrent chemoradiation therapy resulted in lower aHRs (radiation therapy: 0.21; 95% CI=0.04–1.08; p=0.047; concurrent chemoradiation therapy: 0.18; 95% CI=0.03–0.90; p=0.029) compared to patients who received no postoperative treatment. When comparing patients who received treatment, there was no significant difference between those who received concurrent chemoradiation therapy and those who received radiation therapy alone (concurrent chemoradiation therapy compared with radiation therapy, aHR=0.84; 95% CI=0.11–6.30; p=0.861). In group 1, none of the treatments had statistically significant aHR or p values when compared with the group of patients who did not receive any treatment (p=0.248 for radiation therapy and 0.13 for concurrent chemoradiation therapy) (Table 2).

Fig. 3. DFS by adjuvant therapy type within each risk group. Kaplan-Meier curves showing DFS by type of postoperative adjuvant therapy within each group. Group 1: no significant difference by adjuvant therapy (p=0.280); group 2: significant DFS difference between therapy groups (p=0.030); group 3: significant DFS difference between therapy groups (p=0.021).

CCRT, concurrent chemoradiation therapy; DFS, disease-free survival; RT, radiation therapy.

Table 2. Association between postoperative therapy and disease-free survival for each group.

Modality		HR (95% CI)	p-value
Group 1
	None	Reference
	RT	0.53 (0.18–1.58)	0.248
	CCRT	0.31 (0.07–1.43)	0.130
Group 2
	None	Reference
	RT	0.09 (0.01–0.80)	0.031
	CCRT	0.57 (0.15–2.19)	0.409
Group 3
	None	Reference
	RT	0.21 (0.04–1.08)	0.047
	CCRT	0.18 (0.03–0.90)	0.029

CCRT, concurrent chemoradiation therapy; CI, confidence interval; HR, hazard ratio; RT, radiation therapy.

DISCUSSION

We found that the newly proposed JGOG prognostic model for early-stage intermediate-risk cervical cancer has prognostic value in the Korean population. However, unlike the findings from Japan, our study revealed that adjuvant treatment benefited higher-risk patients, specifically those in groups 2 and 3. Therefore, we believe that this risk scoring system can serve as a predictive tool for guiding adjuvant treatment.

The risk grouping system is one of several efforts to update the traditional Sedlis criteria, which were established in the 1990s and have remained unchanged despite significant advancements in imaging modalities and surgical techniques. In particular, the prognostic significance of nonsquamous histology [8,9,10,11], which accounts for up to 32% of all cervical cancers, was excluded from the GOG studies [12], highlighting the need for a more comprehensive risk classification system. Our data support the inclusion of histologic subtypes in the risk grouping system, aligning with the previous risk scoring system developed by the KGOG for identifying the intermediate-risk group [13].

An important methodological difference that may account for the discrepant treatment effect in group 2 between the JGOG model and our cohort is the definition of stromal invasion. The JGOG prognostic model dichotomized invasion into inner versus outer half, whereas our risk stratification retained the Sedlis-derived trichotomy (superficial, middle, deep) that also underpins current National Comprehensive Cancer Network guidance. Consequently, our group 2 was likely enriched for tumors with unequivocally deep stromal infiltration, conferring a higher baseline risk of recurrence and thereby amplifying the observed DFS benefit with adjuvant radiotherapy. These definitional differences in stromal invasion provide a biologically and statistically coherent explanation for the cross-cohort variation in estimated adjuvant treatment effects.

The primary role of risk grouping in early-stage cervical cancer is to identify patients who will benefit from adjuvant treatment. Traditionally, patients classified into the intermediate-risk group are believed to benefit from adjuvant therapy. However, this concept has recently been challenged by several observational studies. A meta-analysis reported no difference in the relative risk of recurrence regardless of adjuvant therapy [14,15,16,17,18,19,20]. The results of some studies suggested that radical surgery alone achieved 5-year DFS rates comparable to those of adjuvant radiotherapy in the risk group [17,18]. Building on these claims and reports, a phase III randomized trial called CERVANTES has recently been planned to validate this evidence [16]. Additionally, the results of several retrospective studies have suggested that concurrent chemoradiation does not offer superior outcomes compared with radiation therapy alone [21,22,23], and a phase III randomized controlled trial focused on this issue, GOG-263 (NCT01101451), is also ongoing. In the Japanese population, this risk grouping was able to aid in providing prognoses model but failed to identify any group that benefited from adjuvant treatment. In contrast, our study was able to identify specific groups that would benefit from adjuvant treatment.

Although the choice of treatment modality may vary globally due to differences in accessibility, institutional resources, and cost, it is crucial to clearly define the subgroup of patients who truly benefit from adjuvant radiation therapy. The aim of this study is not to maximize adjuvant treatment but to optimize it, ultimately guiding the delivery of the minimal yet essential therapy that can be feasibly implemented in diverse clinical settings worldwide. Therefore, this model is also well-suited for application in Western or low-resource settings, where a streamlined and evidence-based approach to adjuvant treatment is particularly valuable.

In the current study, only 2.9% of the patients received postoperative systemic chemotherapy, which was the least common treatment option. In contrast, in a study in Japan, systemic chemotherapy was the most common treatment administered, followed by concurrent chemoradiation therapy, and then radiation therapy, resulting in a greater number of patients receiving chemotherapy. Patients who received adjuvant chemotherapy alone were excluded from the analysis because the current guidelines recommend radiation therapy as a standard adjuvant treatment for the patients included in this study, and the small sample size (2.9% of all patients), which may lead to differences in survival outcomes [24].

This study had several limitations. The study is retrospective, and full information on patients' underlying comorbidities or performance status, as well as the individual surgical details necessary for deciding on each postoperative treatment modality, was not available. Moreover, while the original JGOG study included only Japanese women, the present study was limited to Korean women. As such, differences in tumor biology, genetic background, and healthcare systems between the 2 populations may exist. These differences may limit the generalizability of the findings from both studies to other countries or ethnic groups. In addition, treatment modality selection may vary by institution and physician preference, potentially affecting survival outcomes. Therefore, selection bias related to treatment choice cannot be entirely excluded. Furthermore, as this was a multi-institutional study without central pathology review, discrepancies in pathological assessment may have occurred.

In both this study and the JGOG study, histologic classification was limited to distinguishing between SCC and non-SCC types for risk stratification. Therefore, it is difficult to evaluate oncologic outcomes by specific histological subtypes. Further studies are needed to investigate whether oncologic outcomes differ according to histologic subtype.

In conclusion, this study revealed the prognostic value of histology-incorporated intermediate-risk stratification for early-stage cervical cancer in the Korean population. Unlike findings from the Japanese population, adjuvant treatment benefits high-risk patients. Therefore, this risk scoring system could serve as a guide for recommending adjuvant treatment.