Cervical cancer remains one of the most common malignancies in the world and is still associated with a high risk of death. According to data from the International Cancer Research Institute and the World Health Organization, 604,000 new cervical cancer cases were diagnosed in 2020, resulting in 342,000 deaths [1, 2]. The cause of cervical cancer is strongly related to persistent human papillomavirus (HPV) infection, particularly HPV-16 and HPV-18, which together account for the most cases of invasive disease [3]. Early detection of cervical cancer commonly starts with Pap smear screening, followed by biopsy for confirmation and imaging studies for staging [4]. However, despite improvements in screening and HPV vaccination, many patients are still diagnosed at locally advanced stages, indicating that reliable pretreatment markers that can guide clinical decisions are still needed [5] and that cervical cancer remains a major cause of death. When surgery is not an option—often because the disease is too advanced or the patient is not a suitable surgical candidate—the main treatment strategy shifts to concurrent chemoradiotherapy (CCRT). The definitive radiotherapy (RT) approach typically starts with external beam radiation therapy (EBRT) delivered to the pelvis and is then followed by intracavitary brachytherapy (ICR) to boost the dose directly to the tumor while minimizing exposure to the surrounding healthy tissues [6, 7]. When assessing the prognosis of patients with cervical cancer treated with RT, several pretreatment factors are already known to be important. These can be grouped into a few broad categories that clinicians commonly evaluate before starting treatment. The clinicopathological factors included age, body mass index (BMI), clinical stage, and pathology. The hematological and radiological factors, including hemoglobin (Hgb), hematocrit (Hct), neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), squamous cell carcinoma antigen (SCC-Ag), primary tumor volume (pTV) or size, are also routinely evaluated. In addition, metabolic factors, including the maximum standardized uptake value (SUV_max) in fluorine-18-fluorodeoxyglucose positron emission tomography/computed tomography (F-18 FDG PET/CT) and, when available, metabolic tumor volume (MTV) and total lesion glycolysis (TLG), provide further insight into the overall disease burden [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19]. Despite such well-known pretreatment clinicopathological, hematological, radiological, and metabolic factors, their relative prognostic value has not been systematically compared, and no simple model integrating these factors has been established. Therefore, this study aimed to evaluate the prognostic significance of major and readily obtainable pretreatment factors and to develop a practical, clinically applicable risk-prediction score for cervical cancer patients undergoing CCRT.

The baseline characteristics of the 51 patients are presented in Table 1. The Shapiro–Wilk test results for all continuous variables are provided in Supplementary Table 1. The median age was 58 years (range, 37–81 years), and 84.3% of patients were younger than 75 years. Most patients (72.5%) had FIGO stage III or IV disease, and all underwent CCRT.

The median follow-up duration was 74.6 months (range, 9.9–156.4 months). The median age of the cohort was 58 years (range: 37–81 years). The 5-year OS and PFS outcomes were 75.5% and 72.6%, respectively (Fig. 1).

Fig. 1.

Kaplan–Meier plots of progression-free survival (PFS) and overall survival (OS).

3.1 PFS Analysis

In the univariate analysis, age (p = 0.202), pathology (p = 0.843), Hgb (p = 0.166), NLR (p = 0.210), PLR (p = 0.552), SCC-Ag (p = 0.444), pTV (p = 0.435), and SUV_max (p = 0.342) were not associated with PFS. In contrast, BMI (p = 0.020) and FIGO stage (p 0.001) were associated with PFS (Table 2). According to the multivariate analysis, FIGO stage (HR: 4.385; 95% CI: 1.865–10.310; p 0.001) was a significant prognostic factor for PFS (Fig. 2) (Table 3).

Fig. 2.

Kaplan–Meier survival curves for progression-free survival stratified by factor (FIGO stage), demonstrating statistically significant differences.

Table 2. Univariate Log-Rank analysis of progression-free and OS in cervical cancer patients treated with chemoradiotherapy.

Variable		Univariate analysis
		PFS		OS
		5-year PFS (95% CI)	p-value*	5-year OS (95% CI)	p-value*
Age (years)
	75	69.0% (52.6%–80.7%)	0.202	73.2% (56.8%–84.2%)	0.284
	$\ge$75	87.5% (38.7%–98.1%)		87.5% (38.7%–98.1%)
Pathology
	SCC	72.4% (57.2%–83.0%)	0.843	76.8% (62.0%–86.4%)	0.689
	ADC	66.6% (54.1%–94.5%)		66.7% (54.1%–94.5%)
FIGO stage
	I	100%	0.001	100%	0.001
	II	90.9% (50.8%–98.7%)		90.9% (53.9%–98.8%)
	III	73.6% (54.0%–85.8%)		80.0% (60.8%–90.5%)
	IV	16.7% (7.7%–51.7%)		16.7% (7.7%–51.7%)
BMI (kg/m2)
	21.2	64.7% (37.7%–82.3%)	0.020	51.3% (25.7%–72.1%)	0.024
	$\ge$21.2	85.0% (67.6%–93.5%)		87.7% (70.4%–95.2%)
NLR
	2.2	83.0% (55.9%–94.2%)	0.210	83.0% (55.9%–94.2%)	0.399
	$\ge$2.2	66.0% (47.0%–79.5%)		71.5% (52.3%–84.0%)
PLR
	170.0	73.9% (52.9%–86.6%)	0.552	73.6% (52.4%–86.4%)	0.470
	$\ge$170.0	69.8% (46.9%–84.3%)		78.0% (55.0%–90.2%)
Hgb (g/dL)
	10.2	58.8% (32.5%–77.8%)	0.166	58.8% (32.5%–77.8%)	0.044
	$\ge$10.2	78.8% (60.6%–89.3%)		84.5% (66.6%–93.2%)
SUVmax
	10.1	72.8% (49.0%–86.8%)	0.342	76.6% (52.5%–89.5%)	0.269
	$\ge$10.1	71.2% (50.6%–84.5%)		74.7% (54.1%–87.1%)
SCC-Ag
	27.5	79.2% (61.2%–89.5%)	0.444	81.8% (63.9%–91.4%)	0.231
	$\ge$27.5	57.0% (30.3%–76.8%)		62.5% (34.9%–81.1%)
pTV (cm3)
	35.3	80.0% (40.9%–94.6%)	0.435	90.0% (47.3%–98.5%)	0.188
	$\ge$35.3	70.1% (53.3%–81.8%)		72.0% (55.2%–83.4%)

^* Log-rank test.

Table 3. Summary of prognostic factors identified by multivariate Cox proportional hazards regression for PFS and OS, including the $\beta$-coefficient and risk scoring algorithm.

Variables	Risk factors	HR (95% CI)	$\beta$-coefficient	Risk score	p*
PFS
FIGO stage	I, II, III vs. IV (ref = I, II, III)	4.385 (1.865–10.310)	1.478	14	0.001
BMI (kg/m2)	$\ge$21.2 vs. 21.2 (ref = $\ge$21.2)	2.575 (0.9745–6.805)	0.946	0	0.056
OS
BMI (kg/m2)	$\ge$21.2 vs. 21.2 (ref = $\ge$21.2)	3.207 (1.157–8.893)	1.165	11	0.025
FIGO stage	I, II, III vs. IV (ref = I, II, III)	3.604 (1.559–8.334)	1.282	12	0.003
Hgb (g/dL)	$\ge$10.2 vs. 10.2 (ref = $\ge$10.2)	2.366 (0.858–6.526)	0.861	0	0.096

HR, hazard ratio; ^*Cox proportional hazards regression.

3.2 OS Analysis

Age (p = 0.284), pathology (p = 0.689), NLR (p = 0.399), PLR (p = 0.470), SCC-Ag (p = 0.231), pTV (p = 0.188), and SUV_max (p = 0.269) were not correlated, while Hgb (p = 0.044), BMI (p = 0.024), and FIGO stage (p 0.001) were correlated (Table 2). In the multivariate analysis, BMI (HR: 3.207; 95% CI: 1.157–8.893 p = 0.025), and FIGO stage (HR: 3.604; 95% CI: 1.559–8.334; p = 0.003) showed prognostic significance for OS (Fig. 3) (Table 3). As BMI was the only continuous variable significantly associated with OS, we performed an ROC analysis to determine its optimal cut-off value. The ROC curve for BMI is presented in Supplementary Fig. 1.

Fig. 3.

Kaplan–Meier survival curves for OS stratified by factors (BMI and FIGO stage) demonstrating statistically significant differences.

The results of the univariate analyses for PFS and OS of all factors are summarized in Table 2. A summary of significant and non-significant prognostic factors for both PFS and OS is provided in Supplementary Table 2.

3.3 Risk Stratification Model Based on Prognostic Scoring

A prognostic risk score for PFS was developed based on the $\beta$-coefficients of the risk factors. In this scoring model applied for PFS, FIGO stage (I, II, III, or IV) was incorporated as a prognostic variable, with points assigned based on stage. FIGO stages I–III were allocated a point of 0, whereas FIGO stage IV was allocated a point of 14. Risk stratification for PFS was categorized as follows: low-risk group (point = 0; n = 45, 88.2%), high-risk group (point = 14; n = 6, 11.8%). The two groups showed a significant difference in 5-year PFS (79.5% [95% CI: 64.20%–88.80%] vs. 16.7% [95% CI: 0.77%–51.70%], p 0.001) (Fig. 4). The prognostic risk score for OS was constructed using the following prognostic factors, with weights assigned accordingly: BMI 21.2 kg/m² (assigned a point of 11), and FIGO stage I, II, III, or IV (assigned a point of 0 for stage I–III and 12 for stage IV).

Fig. 4.

Kaplan–Meier plots and calibration curves of (A) PFS and (B) OS according to risk stratification. (A) Kaplan–Meier plot and calibration curve for PFS. (B) Kaplan–Meier plot and calibration curve for OS. The p-values of the Kaplan–Meier survival curves for PFS and OS for risk stratification were p 0.001 and p 0.001, respectively. The prognostic performance of the models was evaluated by comparing predicted probabilities with observed outcomes using calibration plots. The left panel illustrates calibration for PFS, while the right panel shows OS. In both plots, the dashed gray diagonal line represents perfect calibration, where predicted probabilities align perfectly with observed outcomes. The brown curves represent the apparent calibration, which shows how the model performed on its training data. The bias-corrected calibration (shown in blue) was performed using bootstrap resampling (B = 400) to account for potential overfitting. Black ticks at the top of each plot denote the distribution of predicted survival probabilities. The calibration curves for both the PFS and OS models demonstrated good agreement between the predicted and observed probabilities, closely tracking the ideal diagonal line (especially for OS). Deviations in the lower probability range were noted for PFS, possibly due to sparse data in that region. The proximity of the bias-corrected curves to the apparent calibration lines suggested minimal overfitting and good internal validity of the models. Because the number of patients in the high-risk group was small (n = 6 for PFS; n = 3 for OS), both the PFS and OS estimates are associated with wide confidence intervals and should be interpreted with caution as imprecise estimates driven by sparse data.

OS risk stratification resulted in three groups: low-risk group (point = 0, n = 31, 60.8%), intermediate-risk group (point = 11, n = 17, 33.3%), and high-risk group (point = 23, n = 3, 5.9%). A statistically significant differences in the 5-year OS rates was found across the three groups: 93.2% [95% CI: 75.7% to 98.3%] for low-risk group, 57.8% [95% CI: 31.1% to 77.3%] for intermediate-risk group, and 0% [95% CI: NA] for high-risk group p 0.001) (Fig. 4).

Table 3 summarizes the multivariate analysis results for all factors related to PFS and OS, along with the resulting risk stratification scoring model.

In our study, BMI and FIGO stage were significant factors for survival outcomes among patients receiving CCRT for cervical cancer. In our study, the prognostic impact of FIGO stage and BMI was clearly demonstrated by the observed survival differences. Patients with stage I–III had a 5-year PFS of 79.5% and a 5-year OS of 93.2%, whereas those with stage IV showed markedly poorer outcomes, with both 5-year PFS and OS at 16.7%. BMI also showed a significant association with OS, with patients in the higher BMI group achieving a 5-year OS of 87.7% compared with 51.3% in the lower BMI group. Regarding radiotherapy, there was a wide EBRT dose range; however, the higher doses of 84.9 Gy and 70.4 Gy were given only to two patients with bulky pelvic nodal disease, for whom escalation was clinically appropriate while respecting OAR constraints. All other patients received EBRT within the standard range of 50.4–66.6 Gy, and no toxicities were observed in those treated with higher doses. Because all patients subsequently underwent HDR brachytherapy, the overall biological dose to the cervix remained relatively consistent, making it unlikely that these dose variations influenced the results of the study.

Previous study has investigated the relationship between BMI and survival outcomes in gynecological cancers. While obesity has traditionally been considered a negative prognostic factor, recent systematic reviews and meta-analyses have shown that a high BMI does not necessarily confer a survival benefit, and in some cases may even be associated with worse overall survival [14]. Additionally, a recent study have also shown favorable outcomes in higher BMI groups; for example, overweight patients demonstrated significantly better survival than both underweight and obese patients [23]. Furthermore, a qualitative study found that some overweight patients reported better tolerance to therapy, especially RT [24]. In another study, cachexia risk stratification using BMI showed shorter survival (median 4.83 months) in the low BMI groups [25]. Consistent with the studies mentioned above, in our study, BMI was significant for PFS and also for OS. The reasons why BMI might be linked to survival prognosis could be explained as follows: nutritional reserves could buffer against treatment-related cachexia [26], altered pharmacokinetics can lead to improved drug tolerance [27, 28], earlier detection due to more frequent screening could be performed in some regions [29], hormonal milieu possibly affects tumor growth dynamics differently [30], and sarcopenia, rather than BMI alone, serves as a more decisive prognostic factor for OS. Patients with higher BMI and preserved muscle mass tend to exhibit more favorable clinical outcomes [28, 31].

In this study, prognostic factor analysis was performed on patients classified according to the 2018 FIGO staging system. Due to the limited number of patients, the FIGO stages were divided into four groups (I, II, III, and IV) for analytical purposes. When FIGO stage was dichotomized (I–III vs. IV), the prognostic significance remained unchanged, further supporting the robustness of our findings. The results confirmed that the FIGO stage was a statistically significant prognostic factor for both OS and PFS. The 2018 revision of the FIGO staging system, which incorporates imaging and nodal status, is widely recognized as a major advance in predicting cancer outcomes [20, 32]. Studies have demonstrated that these changes allow for earlier intervention, more accurate risk stratification, and tailored treatment, all of which improve outcomes [33, 34, 35]. Overall, these results highlight the complementary value of BMI and FIGO stage as pretreatment prognostic markers. Because our study period included years before the introduction of the 2018 FIGO system, earlier cases were reclassified according to the updated criteria to maintain consistent staging. Since no patient underwent surgery, and the change in staging did not influence the CCRT approach, this reclassification is unlikely to have affected clinical management or the validity of the prognostic analyses. FIGO stage remained the most influential factor, reflecting the overall extent of disease. BMI provided additional insight into treatment tolerance and general health, with lower BMI linked to poorer survival. Because both measures are simple to obtain and represent different aspects of the patient’s status, they offer practical utility in estimating prognosis and can support a straightforward, clinically useful risk-stratification model.

In the present study, hematological factors were not identified as a significant prognostic factor for survival. The prognosis of cervical cancer is significantly affected by anemia, which is commonly indicated by low Hgb and Hct levels. Numerous studies have confirmed that pretreatment anemia correlates with poor tumor oxygenation, leading to reduced radiotherapy efficacy, increased tumor hypoxia, and shorter overall and disease-free survival [11, 12, 36]. These results underscore the clinical importance of monitoring and managing anemia both prior to and during treatment to optimize patient outcomes. In the present study, Hgb demonstrated a significant association with OS in the univariate analysis. In the multivariate analysis, it showed a marginally significant association (HR: 2.366; 95% CI: 0.8578–6.5260; p = 0.096). The multivariate Cox analysis yielded a hazard ratio of 2.366, indicating more than a twofold increased risk of mortality associated with this factor. With a larger cohort, the confidence interval would be expected to narrow, and the observed trend might reach conventional levels of statistical significance. Furthermore, although NLR, PLR, and SCC-Ag have been previously reported as significant prognostic indicators for survival, they were not identified as independent prognostic factors in the multivariate analysis of the present study. Consistent with the findings of the present study, in some studies, NLR, PLR, and SCC-Ag have shown statistical significance in the univariate analyses; however, they were not identified as independent prognostic factors for survival in multivariate analyses [37, 38]. These findings were consistent with the results of the present study and suggested that the interactions among multiple variables should be considered in prognostic evaluations.

While the literature on cervical cancer and age as a prognostic factor is mixed, a study has suggested that advanced age may be an independent negative prognostic factor for mortality [15], whereas others have indicated no significant age-related differences in survival outcomes [8, 39]. Some studies have suggested that younger women may have a higher risk of local recurrence; however, unlike older women, they do not experience worsening outcomes due to recurrence [40, 41]. In our study, age did not have a significant effect on the survival outcomes.

In the present study, prognostic analyses were conducted using pTV as the radiological factor and SUV_max as the metabolic factor. Both factors lacked statistical significance in predicting survival outcomes. Consistent with our findings, other studies consistently reported that pTV was not a statistically significant predictor of survival [13, 16]. While previous studies have demonstrated that the tumor volume reduction rate (TVRR) during treatment is a significant prognostic factor for survival, calculating TVRR requires imaging obtained during radiotherapy, such as CT or MRI [13, 16]. Because the present study aimed to evaluate only pretreatment prognostic factors, TVRR was not included in our analysis. For the same reason, MTV, which also requires post-baseline imaging data, was not incorporated into this study. Yoo et al. [42] reported that SUV_max and average SUV were not independent prognostic indicators for survival. Although a correlation was observed between tumor size and SUV_max, multivariate analysis revealed no direct association with survival. Similarly, several other studies have demonstrated that SUV_max is not significantly associated with survival in multivariate systems and shows only a borderline relationship with OS, indicating that it is not an independent prognostic factor [10, 43]. Although metabolic prognostic indicators, such as MTV and TLG, which reflect the total metabolic burden, have been reported to demonstrate greater prognostic value than SUV_max [44, 45], they are more complex and difficult to measure in routine clinical practice. Therefore, this study included only the SUV_max, which is widely measured and used in most hospitals. In our analysis, radiological and metabolic factors such as pTV and SUV_max were evaluated, but did not show independent prognostic value. This suggests that patient-related systemic factors may exert a greater influence on outcomes than tumor-based measurements. The limited independent significance of pTV and SUV_max appears to reflect their strong correlation with FIGO stage, which already incorporates major elements of tumor burden and nodal involvement. When analyzed with the FIGO stage, much of their prognostic contribution was absorbed by the stage variable. Although SUV_max is less informative than volumetric PET factors such as MTV or TLG, these measures could not be assessed because the required imaging data were unavailable. Nevertheless, SUV_max is widely reported and routinely used in clinical practice, making its inclusion as a pretreatment metabolic factor reasonable in settings where volumetric PET metrics are not accessible. The results of our multivariate analysis using all these factors confirmed that BMI and the FIGO stage were significantly associated with survival. Based on these results, we developed a risk-score stratification model. Although FIGO staging requires assessment through imaging studies, other factors, such as BMI, can be readily obtained in clinical settings, suggesting that this classification model is easily feasible and can be applied to guide treatment decisions. One of the strengths of this study is that it identifies pretreatment prognostic factors that are clinically relevant in routine practice. Building on these findings, we developed a risk-score classification model intended to aid clinicians in planning treatment. To the best of our knowledge, no previous study has integrated multiple established pretreatment prognostic factors into a single predictive model for cervical cancer. By consolidating these factors into a unified framework, this study provides a structured approach that may support more consistent and informed patient assessment before therapy.

In this study, FIGO stage and BMI were the most influential pretreatment factors for survival in patients receiving CCRT for cervical cancer. Their independent prognostic value offers a practical basis for a straightforward risk-stratification approach that may help clinicians recognize higher-risk patients and make more informed treatment decisions. Because both factors are routinely assessed before therapy, they provide a useful means of estimating prognosis at the time of diagnosis. This tool is intended solely to assist with baseline risk communication, not to guide treatment decisions. Although the model shows promise, its clinical reliability will need to be confirmed through external validation in multicenter studies and in larger patient cohorts. These findings provide a sound basis for future work to develop dependable pretreatment prognostic tools.

The datasets analyzed during the current study are available from the corresponding author on reasonable request.

SL and BO were responsible for the conceptualization and design of the study and performed the statistical analysis. Data collection was carried out by SL and HL. The interpretation of the results was contributed by HL, KL, and YK. All authors contributed to critical revision of the manuscript for important intellectual content. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

This Institutional Review Board of Gachon University Gil Medical Center (IRB approval number: GBIRB2025-302) approved this study, which was conducted in compliance with the Declaration of Helsinki. Because this retrospective study used only de-identified clinical data that had already been collected as part of routine care prior to the initiation of the study, no additional patient contact was required; therefore, the requirement for informed consent was waived.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/CEOG47038.