1. Overall Cohort Characteristics
A total of 2936 eligible patients diagnosed with the 2009 FIGO stage I-III cervical cancer and presenting LNM were included in this study. Among them, 73.91% of the patients underwent surgery plus PORT, while definitive RT was administered to 26.09% of the patients. The clinical and pathological data of enrolled patients stratified by the different treatment groups are summarised in Table 1. The results revealed significant disparities in age, marital status, histological type, grade, FIGO stage, tumor size, number of PLNs, and chemotherapy between the two treatment groups (P < 0.05). The squamous cell carcinoma accounted for 81.59% in the definitive RT group, whereas it was 63.32% in the surgery plus PORT group. For the selection of different treatment modalities, patients with stage III were more inclined to undergo definitive RT (19.19% vs. 4.56%). Moreover, in the definitive RT group, tumors larger than 4cm constituted a significantly higher proportion (72.98%) compared to the surgery plus PORT group (45.16%). Additionally, the incidence of number of PLNs ≤ 5 was lower in the definitive RT group than in the surgery plus PORT group (82.11% vs. 90.97%).
Table 1
Characteristics of the patients in this study
| | Definitive RT | Surgery plus PORT | P* |
Age (years) | | | | 0.019 |
| < 45 years | 382 (49.87%) | 1191 (54.88%) | |
| 45–60 years | 284 (37.08%) | 685 (31.57%) | |
| > 60 years | 100 (13.05%) | 294 (13.55%) | |
Race | | | | 0.235 |
| White | 612 (79.89%) | 1720 (79.26%) | |
| Black | 71 (9.27%) | 173 (7.97%) | |
| Others§ | 83 (10.84%) | 277 (12.77%) | |
Marital status | | | | < 0.001 |
| Never married | 253 (33.03%) | 591 (27.24%) | |
| Used married | 156 (20.37%) | 416 (19.17%) | |
| Married | 328 (42.82%) | 1112 (51.24%) | |
| Unknown | 29 (3.78%) | 51 (2.35%) | |
Histology type | | | | < 0.001 |
| SCC | 625 (81.59%) | 1374 (63.32%) | |
| ADC | 84 (10.97%) | 543 (25.02%) | |
| ASC | 37 (4.83%) | 199 (9.17%) | |
| Others | 20 (2.61%) | 54 (2.49%) | |
Grade† | | | | < 0.001 |
| Grade I | 23 (3.01%) | 92 (4.24%) | |
| Grade II | 223 (29.11%) | 817 (37.65%) | |
| Grade III | 300 (39.16%) | 1050 (48.39%) | |
| Grade IV | 20 (2.61%) | 87 (4.01%) | |
| Unknown | 200 (26.11%) | 124 (5.71%) | |
Number of PLNs | | | < 0.001 |
| ≤ 5 | 629 (82.11%) | 1974 (90.97%) | |
| > 5 | 137 (17.89%) | 196 (9.03%) | |
FIGO stage | | | | < 0.001 |
| I | 13 (1.69%) | 46 (2.12%) | |
| II | 606 (79.12%) | 2025 (93.32%) | |
| III | 147 (19.19%) | 99 (4.56%) | |
Tumor size | | | | < 0.001 |
| < 2 cm | 49 (6.39%) | 286 (13.18%) | |
| 2–4 cm | 158 (20.63%) | 904 (41.66%) | |
| > 4 cm | 559 (72.98%) | 980 (45.16%) | |
Chemotherapy | | | | < 0.001 |
| Received | 667 (87.08%) | 1657 (76.36%) | |
| No/Unknown | 99 (12.92%) | 513 (23.64%) | |
Abbreviations: SCC = squamous cell carcinoma, AC = adenocarcinoma. * P was made by χ2-test. |
§ Other = American Indian/AK Native, and Asian/Pacific Islander and Unknown. † Grade I = well differentiated; grade II = moderately differentiated; grade III = poorly differentiated; grade IV = undifferentiated; anaplastic. |
2. Survival Difference Analysis of Primary Treatment
Kaplan-Meier survival analysis revealed that patients who underwent definitive RT exhibited a significantly inferior CSS rate compared to those who received surgery plus PORT (5-year cancer-specific survival rate [5-YCSR]: 70.90% vs. 53.70%, P < 0.001, Fig. 2A). Furthermore, the definitive RT group demonstrated a markedly lower OS rate as well (5-year overall survival rate [5-YSR] 69.20% vs. 50.40%, P < 0.001, Fig. 2B).
To further investigate the impact of different treatment methods and clinicopathological characteristics on survival, we conducted Cox proportional hazard model analysis on overall patients. The results for CSS and OS are presented in Table 2and Supplementary Table 1, respectively. Our findings demonstrated that definitive RT group exhibited a risk of adverse CSS compared to surgery plus PORT group (HR = 0.614, 95% CI: 0.525–0.719, P < 0.001, Table 2). Similarly, receiving definitive RT was associated with an increased risk of poor OS (HR = 0.592, 95%CI: 0.510–0.687, P < 0.001, Supplementary Table 1). Additionally, the COX analysis results indicated that in patients with LNM, the 2009 FIGO stage served as an independent prognostic factor. This suggests that the stage not only reflects patient survival but also influences treatment modality options. Therefore, after stratifying patients according to the 2009 FIGO stage, Kaplan-Meier survival analysis revealed that for FIGO I-II patients, surgery plus PORT group resulted in better CSS (P < 0.001, Fig. 3A-B) and OS (P < 0.001, Fig. 3D-E) compared to definitive RT group. However, for FIGO stage III patients, there was no statistical difference observed between both groups in terms of CSS (42.60% vs. 41.10%; P = 0.465; Fig. 3C) and OS (41.40% vs.34 .80%; P = 0 .083; Fig. 3F).
Table 2
Cox proportional hazard model for prognostic factors in CSS
| | Univariate Analysis | Multivariate Analysis |
| | HR (95%CI) | P | HR (95%CI) | P* |
Treatment | | | < 0.001 | | < 0.001 |
| Definitive RT | 1 | | 1 | |
| Surgery plus PORT | 0.531(0.464–0.608) | | 0.614(0.525–0.719) | |
Age | | | < 0.001 | | 0.338 |
| < 45 years | 1 | | 1 | |
| 45–60 years | 1.270(1.104–1.462) | | 1.086(0.938–1.257) | |
| > 60 years | 1.434(1.187–1.733) | | 1.142(0.936–1.393) | |
Race | | | 0.026 | | 0.029 |
| White | 1 | | 1 | |
| Black | 1.040(0.849–1.273) | | 1.330(1.075–1.647) | |
| Others§ | 1.373(1.045–1.803) | | 0.984(0.801–1.207) | |
Marital status | | 0.028 | | 0.288 |
| Never married | 1 | | 1 | |
| Used married | 1.135(0.948–1.358) | | 1.068(0.886–1.287) | |
| Married | 0.891(0.765–1.037) | | 0.913(0.781–1.067) | |
| Unknown | 0.845(0.547–1.305) | | 0.903(0.583–1.399) | |
Histology type | | < 0.001 | | < 0.001 |
| SCC | 1 | | 1 | |
| ADC | 1.395(1.198–1.624) | | 1.723(1.464–2.027) | |
| ASC | 1.282(1.019–1.611) | | 1.388(1.100-1.752) | |
| Others | 2.620(1.908–3.597) | | 2.335(1.672–3.259) | |
Grade† | | | < 0.001 | | 0.008 |
| Grade I | 1 | | 1 | |
| Grade II | 1.092(0.740–1.611) | | 1.270(0.854–1.891) | |
| Grade III | 1.437(0.981–2.105) | | 1.572(1.060–2.331) | |
| Grade IV | 2.392(1.511–3.788) | | 1.875(1.168–3.008) | |
| Unknown | 1.751(1.162–2.638) | | 1.430(0.938–2.182) | |
Number of PLNs | | < 0.001 | | < 0.001 |
| ≤ 5 | 1 | | 1 | |
| > 5 | 1.952(1.645–2.317) | | 1.611(1.352–1.921) | |
FIGO stage | | | < 0.001 | | < 0.001 |
| I | 1 | | 1 | |
| II | 1.885(1.640–2.165) | | 1.529(1.322–1.770) | |
| III | 2.961(2.435–3.602) | | 1.993(1.611–2.466) | |
Tumor size | | | | < 0.001 |
| < 2 cm | 1 | | 1 | |
| 2–4 cm | 1.632(1.219–2.184) | | 1.400(1.044–1.878) | |
| > 4 cm | 2.839(2.149–3.750) | | 1.942(1.458–2.587) | |
Chemotherapy | | 0.183 | | 0.003 |
| No/Unknown | 1 | | 1 | |
| Received | 0.904(0.778–1.049) | | .0793(0.680–0.925) | |
Abbreviations: SCC = squamous cell carcinoma, AC = adenocarcinoma, HR = Hazard ratio, 95%CI = 95% Confidence interval. |
* Multivariate analysis adjusted for Age, Race, Marital status, Histological type, Grade, FIGO stage, Tumour size, Number of PLNs, and Chemotherapy. |
§ Other = American Indian/AK Native, and Asian/Pacific Islander and Unknown. |
† Grade I = well differentiated; grade II = moderately differentiated; grade III = poorly differentiated; grade IV = undifferentiated; anaplastic. |
Moreover, utilizing prognostic risk factors identified through multivariate COX analysis, we stratified the data based on age, number of PLNs, and tumor size to assess the impact of different treatment regimens on survival outcomes. The results found that among patients with number of PLNs ≤ 5, those who underwent surgery plus PORT exhibited significantly improved CSS in contrast to those with definitive RT (5-YCSR: 72.80% vs. 55.80%; P < 0.001; Fig. 4A), as well as OS (5-YSR: 71.20% vs. 52.70%; P < 0.001; Fig. 4C). However, in the subgroup of patients number of PLNs > 5, there was no significant difference between CSS and OS for the two treatment modalities (CSS: P = 0.149, Fig. 4B; OS: P = 0.080, Fig. 4E). Furthermore, when analyzing different age groups and tumor sizes, it was evident that surgery plus PORT yielded significantly better CSS and OS across all subgroups compared to definitive RT (Fig. 5, Supplementary Fig. 1). Therefore, these results suggested that FIGO stage and number of PLNs were the main factors influencing divergent clinical outcomes between the two treatment modalities.
3. Effect of Primary Treatment on Survival outcomes
FIGO stage and number of PLNs are the key factors affecting the prognosis. To accurately assess the impact of treatment modalities on prognosis, multivariate COX analysis was conducted on CSS and OS data, stratified by FIGO stage and number of PLNs. In patients with FIGO stage I-II, those who received surgery plus PORT were equipped with significantly higher CSS (HR 0.544, P < 0.001; HR 0.532, P < 0.001; univariate and multivariate analysis, respectively; Table 3) and OS (HR 0.557, P < 0.001; HR 0.536, P < 0.001; univariate and multivariate analysis, respectively; Supplementary Table 2) compared to those with definitive RT. However, in patients with FIGO stage III, there were no statistically significant differences observed in CSS (HR 0.879, P = 0.470; HR 1.090, P = 0.698; univariate and multivariate analysis, respectively; Table 3) and OS (HR 0.747, P = 0.087; HR 0.747, P = 0.087; univariate and multivariate analysis, respectively; Supplementary Table 2) between the two treatment modalities. Considering the impact of FIGO staging and number of PLNs involvement on treatment decisions, an interaction analysis was conducted to evaluate the treatment effect on survival outcomes in different subgroups. Given the similar survival benefits observed in FIGO stage I and II subgroups, we combined their data for analysis. Our findings revealed a distinctly different interaction effect on CSS of these two treatment modalities between FIGO stage I-II and stage III (univariate P interaction = 0.020, multivariate P interaction = 0.007; Fig. 6). Specifically, surgery plus PORT was identified as a protective factor for CSS in patients with stage I-II. Conversely, no significant interaction effects were observed between different treatment modalities for overall survival (OS) (univariate P interaction = 0.125, multivariate P interaction = 0.051; Supplementary Fig. 2). Furthermore, there were significant differences in the CSS and OS interactions between the two treatment modalities based on number of PLNs > 5 and ≤ 5 within subgroups (CSS: univariate P interaction = 0.020, multivariate P interaction = 0.026; Fig. 6; OS: univariate P interaction = 0.025, multivariate P interaction = 0 .028; Supplementary Fig. 2). In the subgroup with number of PLNs ≤ 5, patients who underwent surgery plus PORT showed superior CSS compared to those with definitive RT (univariate HR 0.517, P < 0.001; multivariate HR 0.587, P < 0.001; Table 3) as well as OS (univariate HR 0.507, P < 0.001; Multivariate HR 0.569, P < 0.001; Supplementary Table 2). However, in patients with number of PLNs > 5, both groups showed no significant difference in survival benefit in respect of CSS (univariate HR 0.792, P = 0.152; Multivariate HR 0.769, P = 0.190; Table 3) and OS (univariate HR 0.765, P = 0.083; Multivariate HR 0.713, P = 0.077; Supplementary Table 2).
Table 3
Stratified analysis of 5-year cancer specific survival in this study
| Subgroup | 5-year cancer specific survival rate (95% CI) | HR (95% CI); P value |
| Surgery plus PORT | Definitive RT | Univariate | Multivariate* |
FIGO stage | | | | |
| I | 0.770(0.746–0.793) | 0.637(0.576–0.691) | 0.544(0.439–0.672) ;<0.001 | 0.532(0.420–0.673) ;<0.001 |
| II | 0.608(0.564–0.649) | 0.481(0.412–0.547) | 0.664(0.535–0.824) ;<0.001 | 0.584(0.459–0.744) ;<0.001 |
| III | 0.426(0.318–0.529) | 0.411(0.321–0.498) | 0.879(0.619–1.248);0.470 | 1.090(0.705–1.687);0.698 |
Number of PLNs | | | | |
| ≤ 5 | 0.728(0.706–0.748) | 0.558(0.513–0.601) | 0.517(0.445–0.601) ;<0.001 | 0.587(0.493–0.699) ;<0.001 |
| > 5 | 0.517(0.438–0.591) | 0.436(0.339–0.530) | 0.792(0.576–1.090) ;0.152 | 0.769(0.519–1.139) ;0.190 |
* Multivariate analysis adjusted for Age, Race, Marital status, Histological type, Grade, FIGO stage, Tumour size, Number of PLNs, and Chemotherapy. |