Introduction
Cervical cancer is the fourth most common cancer among women worldwide, with an estimated 660,000 new cases and 350,000 deaths reported in 2020 (1). Concurrent chemoradiotherapy (CCRT) –the standard treatment for cervical cancer– combines chemotherapy and radiation therapy and has well-established efficacy. Cisplatin (CDDP) is the most widely used anticancer drug in CCRT, with a standard recommended dose of 40 mg/m2 (2, 3).
Although CDDP exhibits potent antitumor effects, its emetogenicity creates a significant challenge. Standard prophylactic antiemetic regimens include four-drug combination therapy comprising an NK1 receptor antagonist (NK1-RA), a 5-HT3 receptor antagonist (5-HT3-RA), olanzapine, and steroids such as dexamethasone (DEX) (4-6).
Although CDDP is classified as highly emetogenic chemotherapy, a single-agent dose of 40 mg/m2 is often classified as moderately emetogenic chemotherapy; however, the emetogenic risk increases with CCRT because of the effects of radiation (7, 8). Therefore, strong antiemetic prophylaxis is necessary for highly emetogenic chemotherapy. The latest guidelines from the Multinational Association of Supportive Care in Cancer and the European Society for Medical Oncology explicitly recommend triple-agent prophylaxis (NK1-RA + 5-HT3-RA + DEX) with weekly CDDP 40 mg/m2 administration (6, 9).
Steroids such as DEX are essential components of this antiemetic prophylaxis protocol; therefore, appropriate prevention of chemotherapy-induced nausea and vomiting (CINV) is recommended because it can severely affect patients’ quality of life and treatment adherence (10, 11).
An increased risk of adverse effects with continuous steroid administration is a critical concern, particularly in the context of CCRT where patients receive repeated weekly doses for 5-6 consecutive weeks. In contrast to intermittent highly emetogenic chemotherapy cycles, the cumulative dose of DEX in CCRT can become substantial, potentially exacerbating dose-dependent toxicities. Although steroids have potent antiemetic effects, they are associated with numerous adverse effects, including hyperglycemia, insomnia, immunosuppression, and adrenal suppression (12-15).
In patients with cervical cancer undergoing CCRT, these risks are further compounded by pelvic irradiation, which induces lymphopenia and intestinal mucosal damage. Therefore, minimizing exogenous steroid exposure is beneficial not only for metabolic health but also for maintaining the patient’s immune status and treatment compliance over prolonged treatment periods (16).
These adverse effects can pose major clinical problems, particularly for patients with comorbidities such as diabetes or cardiovascular disease and patients with cancer with compromised immune function. Therefore, clinical strategies for shortening the DEX administration duration (steroid-sparing therapy) should be established to maintain antiemetic efficacy while reducing the burden of steroid-related adverse effects (17, 18). Previous studies have examined the impact of steroid-sparing therapy on CINV in multiple clinical trials (19-21).
However, evidence regarding the safety and efficacy of CCRT with CDDP 40 mg/m2 –which requires potent antiemetic prophylaxis according to the highly emetogenic chemotherapy guidelines (because of radiation-induced nausea enhancement and prolonged treatment duration)– for cervical cancer remains limited.
We evaluated the clinical feasibility of steroid-sparing therapy comprising CCRT with CDDP 40 mg/m2 for patients with cervical cancer based on the rate of total control (TC) of nausea and vomiting, defined as the absence of nausea and vomiting without rescue medication, as the primary endpoint to avoid the risk of adverse effects associated with repeated DEX administration.
Patients and Methods
Patient selection. This multicenter retrospective study included a post hoc analysis of data collected from 1,080 patients across 14 participating hospitals in Japan, including university hospitals, cancer centers, and general hospitals, who underwent CCRT between January 2016 and March 2024 (22). Patients who were treated with nedaplatin, did not use NK1-RA, had a history of anticancer drug administration, switched from CDDP to nedaplatin during the study period, participated in other clinical trials or interventional studies during the study period, and had missing data were excluded.
All selected patients received CCRT comprising CDDP 40 mg/m2 weekly (Figure 1). Patients were classified into the steroid-sparing (NK1-RA + 5-HT3-RA + DEX on day 1) and non-steroid-sparing (NK1-RA + 5-HT3-RA + DEX days 1-3) groups.
Data regarding age, height, weight, body surface area (BSA), opioid use, total radiation dose, cancer stage, and concomitant medications were collected before the first CCRT cycle. BSA was calculated using the DuBois formula: BSA (m²) = (weight [kg])0.425 × (height [cm])0.725 × 0.007184 (23).
TC was stratified as the absence of vomiting and nausea without rescue antiemetic administration during the acute phase (0-24 h), delayed phase (24-120 h), and the period beyond the delayed phase (120-168 h) following CDDP administration during the initial CCRT cycle, as well as throughout the entire 42-day CCRT period spanning multiple cycles (24).
The TC achievement rate was evaluated according to the Common Terminology Criteria for Adverse Events version 5.0 based on the usage history available in the electronic medical records.
Statistical analysis. Patient characteristics were analyzed using the Fisher’s exact test for categorical variables and the Mann-Whitney U test for continuous variables. The primary endpoint, namely TC achievement during the entire CCRT period, was analyzed using a multivariate logistic regression analysis.
To aggregate the weekly repeated observations into this patient-level endpoint, repeated measurements of complete control (TC) over the entire CCRT course were synthesized into a single binary outcome: patients who maintained complete control (i.e., the absolute absence of vomiting, nausea, and no rescue medication use) across all assessed time points were classified as achieving TC, whereas those who experienced treatment failure in even a single time point (or cycle) were classified as non-TC. Regarding secondary endpoints, numbers and proportions for binary data, as well as 95% confidence intervals (CIs), were calculated according to the data characteristics. Statistical significance was set at p<0.05 for all adjusted analyses, except for the TC achievement rates, which were adjusted using the Bonferroni correction for multiple comparisons (significance level: p<0.0125).
Regarding survival time data, the Kaplan-Meier method was used to estimate survival curves, with TC failure as the event and treatment discontinuation or loss to follow-up as censors. Intergroup differences were analyzed using the log-rank test, and hazard ratios (HRs) with 95% CIs were calculated using a Cox regression analysis.
A Cox regression analysis of the duration of TC maintenance during the entire CCRT period was performed, with age, opioid use, total radiation dose, olanzapine, and 5-HT3-RA as explanatory variables. These potential confounding factors were selected after careful review of the relevant literature (7, 25-28). To further validate the robustness of the primary findings, sensitivity analyses were performed.
First, a propensity score-matching (PSM) analysis was conducted to adjust for imbalances in baseline characteristics between the steroid and non-steroid-sparing groups. The propensity score was calculated using a logistic regression model with the following covariates: age, olanzapine use, total radiation dose, palonosetron use, and opioid administration. Patients were matched in a 1:1 ratio using the nearest-neighbor method with a caliper width of 0.2.
Second, a subgroup analysis was performed focusing on patients who did not receive olanzapine to evaluate the effect of steroid-sparing independent of olanzapine use. Statistical analyses were performed using EZR software version 1.68 (29).
Ethics. This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Research Ethics Committee of the Ehime University Hospital (approval number: 2511005). Informed consent for participation was obtained using an opt-out approach that ensured that all participants were provided with the opportunity to decline participation.
Results
Patient characteristics. The patient selection process is summarized in Figure 1. Of the 1,080 hospitalized patients screened, 501 were excluded: 115 patients who received treatment with nedaplatin, 240 who did not use NK1-RA, 66 with a history of anticancer drug administration, five who switched from CDDP to nedaplatin during the study period, 17 who participated in other clinical trials or interventional studies during the study period, and 58 classified as “other,” including those treated at other hospitals or transferred patients.
Ultimately, 579 patients were included in this analysis: 80 in the steroid-sparing group and 499 in the non-steroid-sparing group.
Baseline patient characteristics are summarized in Table I. The median age of all patients was 56 years. Of the 579 patients, 37 (6.3%) used opioids, specifically oxycodone, fentanyl, or hydromorphone.
No statistically significant differences in age, height, weight, BSA, cancer stage, opioid usage rate, or 5-HT3-RA type were observed between groups. The median total radiation dose significantly differed between groups (p=0.032).
Regarding antiemetic use, 76 (13.1%) of the 579 patients received olanzapine. In the steroid and non-steroid-sparing groups, 28.8% (23/80) and 10.6% (53/499) used olanzapine, respectively; this difference was statistically significant (p<0.001). Cancer stage data of one patient were missing.
Primary endpoint. The TC achievement rates during the entire CCRT period were 30.0% and 36.5% in the steroid and non-steroid-sparing groups, respectively (OR=0.746; 95% CI=0.447-1.250; p=0.263) (Table II).
Additionally, the TC achievement rates in the acute phase were 98.8% and 96.0% in the steroid and non-steroid-sparing groups, respectively (OR=3.300; 95% CI=0.436-24.900; p=0.247). The TC achievement rates in the delayed phase were 60.0% and 71.7% in the steroid and non-steroid-sparing groups, respectively (OR=0.591; 95% CI=0.363-0.962; p=0.035). However, after applying the Bonferroni correction for multiple comparisons (significance threshold: p<0.0125), this difference in the delayed phase was not statistically significant.
The TC achievement rates during the period beyond the delayed phase were 57.5% and 60.9% in the steroid and non-steroid-sparing groups, respectively (OR=0.868; 95% CI=0.538-1.400; p=0.561) (Table II).
The steroid-sparing group had a slight tendency toward a higher occurrence of events compared with the non-steroid-sparing group; however, the difference was not statistically significant (HR=1.205; 95% CI=0.907-1.601; p=0.199) (Figure 2). The median times to TC failure were 4 (interquartile range=3-12) days and 5 (interquartile range=3-14) days in the steroid and non-steroid-sparing groups, respectively.
To exclude the influence of other important confounding factors, such as characteristics with significant differences from those at baseline, a multivariate logistic regression analysis –adjusted for age, opioids, radiation dose, olanzapine, and 5-HT3-RA– was performed. There was no statistically significant difference in the overall TC achievement rate during the entire CCRT period (OR=0.826; 95% CI=0.480-1.420; p=0.490) between the groups (Table III).
Sensitivity analyses. Sensitivity analyses further supported these findings. In the PSM cohort (n=158), adjusted for key covariates including olanzapine use, age, radiation dose, and opioids, no significant difference was observed between the groups (OR=0.736; 95% CI=0.370-1.460; p=0.381).
Furthermore, univariate logistic regression analysis within this PSM cohort confirmed that there was no statistically significant difference in the delayed phase TC rate between the two groups (OR=0.676; 95% CI=0.350-1.310; p=0.243).
Similarly, a subgroup analysis of patients not receiving olanzapine (n=503) showed consistent results (OR=0.833; 95% CI=0.464-1.490; p=0.531) (Table IV).
Discussion
We evaluated the potential of steroid-sparing therapy as prophylactic treatment for nausea and vomiting in patients with cervical cancer undergoing CCRT comprising CDDP 40 mg/m2. To our knowledge, this is the first study to evaluate the effect of steroid-sparing therapy on antiemetic efficacy during CCRT comprising CDDP 40 mg/m2 based on the TC achievement rate using a multicenter, multivariate analysis. After adjustments were performed using a multivariate logistic regression analysis, no significant difference in the TC achievement rate during the entire CCRT period was observed between the steroid and non-steroid-sparing groups (OR=0.826; 95% CI=0.480-1.420; p=0.490).
These results suggest that shortening the duration of steroid administration may not compromise TC in patients receiving CCRT with CDDP regimens. Previous studies demonstrated noninferiority in delayed and overall complete response rates for vomiting when olanzapine or NK1-RA was added to phase III trials involving anticancer regimens containing CDDP (≥50 mg/m2); our results are consistent with these findings (20, 21). Furthermore, recent studies in other emetogenic settings, such as lymphoma regimens, similarly suggest that risk-based or individualized antiemetic optimization is clinically appropriate and effective (30).
Our findings provide important clinical evidence for patients with cervical cancer undergoing CCRT who are concerned about the adverse effects associated with long-term steroid use, such as increased blood glucose, immunosuppression, and osteoporosis.
However, a comparison of the TC achievement rates based on the time revealed an interesting trend. In the delayed phase, the TC achievement rate in the non-steroid-sparing group (71.7%) was 11.7% higher than that in the steroid-sparing group (60.0%), indicating a tendency toward an increased risk of treatment failure (OR=0.591; 95% CI=0.363-0.962; p=0.035). However, this result did not reach significance after corrections for multiple comparison (significance threshold: p<0.0125); therefore, cautious interpretation is necessary.
Crucially, this apparent difference completely disappeared in the propensity score-matched cohort, where background variables were strictly balanced, showing no statistically significant difference in the delayed phase between the two groups (p=0.381). This indicates that the lower TC rate initially observed in the univariate analysis was heavily driven by baseline confounding factors, rather than the steroid reduction itself.
Nevertheless, these results provide biological support of the importance of the anti-inflammatory effects of steroids and their central antiemetic action, particularly controlling delayed CINV, and represent a clinically non-negligible difference (31, 32). Given that gastrointestinal symptoms such as nausea, vomiting, and nutrition impact symptoms heavily influence patient prognosis, nutritional status, and treatment tolerance in oncology, preventing and managing these symptoms is paramount (33). Therefore, when introducing steroid-sparing therapy, considering its possible inferiority to standard therapy for delayed CINV and the subsequent need for either augmentation with alternative agents, such as olanzapine, or rigorous monitoring is essential.
Herein, the significantly higher rate of olanzapine use in the steroid-sparing group compared with that in the non-steroid-sparing group (28.8% vs. 10.6%; p<0.001) warrants further consideration. Since olanzapine is an antiemetic agent, this imbalance suggests the presence of treatment intervention bias involving the intentional or incidental compensation for the lack of steroids with the addition of olanzapine in the steroid-sparing group (34-36).
This pattern has also been observed in our recent real-world investigation of high-risk chemotherapeutic settings (37).
However, because of the retrospective nature of this study, a definitive determination of olanzapine as an active intervention based on physician discretion or an incidental clinical decision made in response to the patient’s condition or adverse effects is difficult. Therefore, the presence of potential confounding factors should be recognized. We incorporated olanzapine into a multivariate logistic regression analysis for adjustment (OR=0.826; p=0.490), which supported the reliability of our primary conclusion.
Furthermore, we performed sensitivity analyses to address this concern. In the PSM cohort, where olanzapine use and other clinical backgrounds were strictly balanced, the results remained consistent (OR=0.736, p=0.381). Additionally, a subgroup analysis focusing on patients who did not receive olanzapine (n=503) also showed no significant difference between the groups (OR=0.833, p=0.531).
These consistent findings across multiple analytical approaches strongly suggest that the antiemetic efficacy of steroid-sparing therapy is independent of olanzapine use, aligning with the broader principle that proactive supportive care management effectively mitigates gastrointestinal toxicities and optimizes treatment tolerability across diverse clinical settings (38).
Another confounding factor that requires further attention is the distribution of the total radiation dose at baseline. The steroid-sparing group had a significantly higher median total radiation dose than that of the non-steroid-sparing group (74 Gy vs. 62 Gy; p=0.032).
Radiation exposure to the pelvic cavity promotes serotonin release via intestinal mucosal damage, leading to radiation-induced nausea and vomiting (39). Increased radiation doses and expanded irradiation fields are risk factors for radiation-induced nausea and vomiting (40). This suggests that the steroid-sparing group may have had a higher potential risk of emetogenicity than the non-steroid-sparing group.
However, a multivariate logistic regression analysis revealed no statistically significant difference in the overall TC achievement rate during the entire CCRT period (OR=0.826; p=0.490). This was further corroborated by our PSM analysis, which was adjusted for total radiation dose as a covariate and revealed no significant difference between the groups. This suggests that steroid-sparing can maintain its antiemetic effect, even after accounting for the potential risk of radiation-induced nausea and vomiting.
Study limitations. First, because this was a retrospective cohort study, unobserved confounding factors (such as the criteria for steroid selection, patient compliance, and variability in subjective nausea assessment) may have existed. Second, the trend of the TC achievement rates during the delayed phase (p=0.035), which was deemed nonsignificant after correction for multiple comparisons, may have been associated with the sample size and cohort characteristics.
Third, the lack of data regarding steroid-related adverse events represents the most serious limitation of this study. The clinical significance of steroid-sparing therapy is based on its ability to maintain antiemetic efficacy while avoiding adverse effects. Herein, steroid-sparing therapy did not exhibit a statistically significant difference in the ability to control the TC achievement rate compared with non-steroid-sparing therapy; however, we did not evaluate its crucial benefit of reducing steroid-related adverse events.
Therefore, we could not directly demonstrate whether the steroid-sparing protocol actually reduced the incidence or severity of these toxicities, which should be explored in future studies.
Conclusion
The results of this study suggest that steroid-sparing therapy may not significantly compromise efficacy in terms of antiemetic effects during CCRT comprising CDDP for patients with cervical cancer. The consistency of the results across multivariate, PSM, and subgroup analyses supports the clinical feasibility of this approach. Further prospective randomized controlled trials are warranted to definitively validate these findings and evaluate the reduction in steroid-related adverse events.
Conflicts of Interest
The Authors have no relevant financial or non-financial interests to disclose.
Authors’ Contributions
Conceptualization: All authors. Data curation: Takuma Matsumoto, Masaki Nakai, Yusuke Kawamura, Yoshinobu Gohara, Naoto Hoshino, Hajime Morita, Naoya Tonomura, Manami Banba, Ayako Yamaguchi, Masaki Tachibana, Tomoki Fukushima, Hiroki Hosokawa, Takuya Mura, Tsuyoshi Yabuki, and Noriaki Hidaka. Formal analysis: Takuma Matsumoto.
Writing – original draft: Takuma Matsumoto. Writing – review & editing: All Authors.
Acknowledgements
The Authors would like to express their sincere gratitude to Natsumi Yamashita of the NHO Shikoku Cancer Center for her advice regarding statistical analysis. They also express their sincere gratitude to all those who cooperated with us during the conducting of this study. The Authors would like to thank Editage (www.editage.jp) for English language editing.
Funding
The Authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
Artificial Intelligence (AI) Disclosure
No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.