Cervical cancer remains the fourth most common and lethal female malignancy worldwide, with an estimated 569,847 new cases and 311,365 deaths worldwide reported in 2018 (1). Currently, the average 5-year survival rate of cervical cancer has reached 66% in developed countries, yet less than half of patients from developing countries could live longer than 5 years (2, 3). Several controversies still exist for the optional management of locally advanced cervical cancer [LACC; International Federation of Gynecology and Obstetrics (FIGO) stages IB2-IVA], which represents almost 60% of all diagnosed cervical cancers, with a 3-year OS about 81% for stage IB2, 51% for stage IIIB and 28% for stage IVA, respectively (4).

Many trials have shown that concurrent chemoradiotherapy (CCRT) reduces the risk of death for LACC by 30 to 50% compared with radiotherapy (RT) alone (5–9). Based on these data, the National Cancer Institute suggested that strong consideration should be given to using CCRT instead of RT alone for LACC (10). Surgery is still a common treatment option, and neoadjuvant chemotherapy (CT) before surgery has been shown to improve survival in selected LACC patients (11, 12). Although the approach of neoadjuvant CT plus surgery lacks adequate evidence, it is practiced in many parts of the world (13, 14). The role of adjuvant CT after CCRT for LACC has also been explored in many studies (15–18). However, there is much debate because four randomized controlled trials of adjuvant CT after CCRT have inconsistent data when compared with CCRT (15–18). Presently, although more interests are focused on neoadjuvant CT before CCRT (19–21), there is only one phase II research about addressing this strategy compared with CCRT (21). With lots of neoadjuvant and adjuvant therapies, the optimal strategy for the management of LACC remains to be characterized.

Network meta-analyses (NMA) provide an opportunity to perform direct and indirect treatment comparisons among randomized studies without breaking randomization, as long as specific assumptions are fulfilled (22). Through indirect measures, NMA enables estimation of comparative efficacy for interventions that have not been investigated in direct head-to-head randomized trials (e.g., comparison of the treatment A vs C, using data from trials comparing A vs B and B vs C). Thus, we employed Bayesian NMA to compare the outcomes of different treatment modalities for LACC.

To our knowledge, this study was the first one to compare all commonly practiced treatment modalities for LACC. We integrated both the direct and indirect comparisons by employing a Bayesian NMA, remedying the insufficiency of traditional meta-analysis. We found that, among nine modalities, CCRT and CCRT plus adjuvant CT were likely to be the optimal strategies for LACC in terms of both 3-year OS and PFS.

Neoadjuvant treatments have the potential to eradicate micrometastases and could reduce systemic failures, in addition to facilitating local control by surgical resection. Neoadjuvant treatments also increase radio-sensitivity and decrease the hypoxic cell fraction. A recent review did not find sufficient evidence that neoadjuvant treatments followed by surgery improved the survival of LACC when compared with patients who were treated with RT or CCRT alone (47). It also failed to distinguish the modality of neoadjuvant treatments, i.e., CT, RT and CCRT. In our NMA, a total of 18 randomized controlled trials about neoadjuvant treatments (i.e., neoadjuvant CT plus surgery, neoadjuvant CT plus CCRT, neoadjuvant RT plus surgery and neoadjuvant CCRT plus surgery) were covered. Still, none of the neoadjuvant treatment outcomes surpassed that of CCRT alone.

In contrast to previous meta-analysis (11), the risk reduction of death associated with neoadjuvant CT was not statistically significant in our NMA. The previous meta-analysis did not include 3 randomized controlled trials by Wen et al. (39), Li et al. (40), and Yang et al. (46). In addition, none of the previous meta-analyses compared neoadjuvant CT followed by surgery with CCRT, the standard treatment modality. There was only one head-to-head phase III clinical trial (29), in which Gupta et al. found it was no difference in OS between these two strategies; however, CCRT resulted in superior disease-free survival in stage IIB disease. Neoadjuvant CT before CCRT was also evaluated in the prospective randomized phase II trial (21), which suggests that the addition of neoadjuvant CT to CCRT is associated with an inferior PFS and a lower OS when compared with CCRT. Nevertheless, the result of one ongoing head-to-head phase III clinical trial evaluating the role of neoadjuvant CT plus CCRT will be of interest (ClinicalTrials.gov identifier: NCT01566240). Although no improvement in survival was found when compared with surgery alone, neoadjuvant CT (48), neoadjuvant RT (37), and neoadjuvant CCRT (31, 32) before surgery are also used in reality and clinical studies for LACC.

A meta-analysis found adjuvant CT after CCRT may be beneficial because 35% of patients experience disease progression after CCRT alone (49). Although many prospective phase II studies show an increased response rate with adjuvant CT after CCRT with high 80–90% survival rates (50–52), there is still much debate when compared with CCRT. Two randomized controlled trials showed increased PFS or OS using CCRT plus adjuvant CT (16, 17) whereas two other trials could not demonstrate such a benefit (15, 18). Overall, our summary analysis suggested that adjuvant CT after CCRT bring a significant PFS but no OS advantage compared with CCRT. Our SUCRA displayed that CCRT plus adjuvant CT was the best therapeutic options in terms of 3-year PFS. The interesting findings could be explained by that the adjuvant CT after CCRT eradicated the micrometastases that may have not be eradicated by CCRT. The adjuvant CT could also consolidate the local effects of the concomitant chemo(brachy)radiotherapy (17). Thus, adjuvant CT after CCRT could be beneficial for certain patient groups. However, toxicities were more frequent in the adjuvant CT after CCRT group. For example, in the phase III trial, grade 3 and 4 toxicities were 86.5% for adjuvant CT after CCRT arm, but 46.3% for CCRT arm (P <0.001) (16). However, in this trial, a possible selective benefit of adjuvant CT on a more advanced stage or certain histopathology (adenocarcinoma in particular) might be diluted by including stage II patients or squamous cell carcinoma, and they comprised the majority cervical cancer population and these diseases could be managed by CCRT alone. In line with this, Dueñas-González et al. (16) also found greater benefit of adjuvant CT after CCRT were observed in stages III–IV or for adenocarcinoma. Therefore, phase III trials specifically targeted for specific disease populations such as stage III–IV disease and those with adenocarcinoma are needed to elucidate whether adjuvant CT after CCRT is better than CCRT under certain circumstances.

Radical hysterectomy with pelvic lymphadenectomy has been a treatment option for LACC (category 2B) (53). In our study, CCRT showed a significant PFS and OS advantage over surgery alone by indirect comparison. Currently, there are no prospective randomized controlled trials directly investigating surgery in compared with CCRT for LACC. Since 20-year OS of RT group and surgery group were 77 and 72% (P = 0.280) (54) and the National Cancer Institute suggested that CCRT should replace RT alone for LACC (10), direct comparison of surgery and CCRT might not be necessary. Thus, in general, our results still support that CCRT is the treatment of choice for LACC. The refinement of CCRT regimens should be considered as the direction for future research.

Our conclusion is strengthened by the Bayesian NMA used. Firstly, this study pooled together direct and indirect evidence within comprehensive comparisons. Secondly, we identified regimens that were superior to others, which made interpretation straightforward from a clinical point of view. The limitations of this NMA also need to be acknowledged. Firstly, we did not have access to individual patient data, which limited the precision of our estimates. Secondly, the characteristics of the included studies confined the quality of our analysis. Although transitivity assumption was met, there was no accepted method to test similarity and subjective judgments can only be made by comparing clinical studies characteristics. Thirdly, the CT, RT, CCRT and surgery techniques and regimens also differ. The studies were reported over a span of 32 years (1987–2019). Thus, the availability of the treatment facilities and technology over the years should be very much different (e.g., RT planning and delivery, CT drugs and FIGO staging). Nevertheless, the current NMA is not a substitute for direct head-to-head comparison trials, but suggests which treatment modality may represent the most appropriate for further evaluation in future studies.

In conclusion, our Bayesian NMA supports CCRT as the standard therapy for LACC. Compared with CCRT, although CCRT plus adjuvant CT has shown a PFS benefit for LACC, future studies needed to find an appropriate chemotherapy regimen which improved OS in some patient groups. Therefore, in general, CCRT is still the treatment of choice for LACC. Current and future research should be focused on developing the most effective CCRT and CCRT plus adjuvant CT regimens.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

BP and YQ designed the study. YQ and HL completed the data collection. YQ and BP drafted the manuscript. BP provided clinical insights and did the literature review and help with the drafting of the manuscript. All authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.