Trial Design

SENTICOL III is an international, randomized, multicenter, single-blind trial. It will compare SLN biopsy alone with SLN biopsy + pelvic lymphadenectomy for disease-free survival and health-related quality of life in patients with negative SLNs (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Trial schema. DFS, disease-free survival; ICG, indocyanine green; OS, overall survival; QOL, quality of life; RFS, recurrence-free survival; SLN, sentinel lymph node.

Patients will undergo SLN biopsy using a strict methodology. For those in whom a radiotracer will be used, the injection will be performed the day before surgery. Long and short protocols are allowed. Pre-operative imaging is required for those who undergo radiotracer mapping, with single photon emission CT combined with CT being the recommended option but with lymphoscintigraphy also acceptable. Laparoscopic, robotic, or open access will be obtained and adhesiolysis will be performed if necessary. The cervical injection can be performed either before or after accessing the abdomen. At least two deep injections (3 and 9 o’clock) are needed (supplementary injections are allowed). Detection can be performed with isotope alone, isotope + blue dye, isotope + indocyanine green or indocyanine green alone. Blue dye alone is not allowed.

The MSKCC algorithm will be followed (figure 2).4 All peritoneal surfaces will be inspected and any suspicious lesions biopsied. If there is extra-cervical spread then SLN mapping is unnecessary and the procedure will be terminated as the patient will require other treatment. The retroperitoneal spaces must be systematically opened and examined. Any suspicious nodes are to be extracted regardless of mapping. The MSKCC criteria must be fulfilled, especially the bilateral detection of SLNs. The SLNs will be systematically analyzed by frozen section: SLNs should be cut in half along their long axis, one level in one of the two parts, staining with hematoxylin-eosin-safran. Patients with negative SLNs in the frozen section will then be randomized 1:1 between the experimental arm (SLN biopsy alone) or the standard arm (SLN biopsy + pelvic lymph node dissection)(figure 1). Patients with unilateral detection or a non-optimal mapping of SLNs will undergo a side-specific lymphadenectomy. Patients with a positive frozen section will not be randomized and will be included included in the pN1 cohort. In these patients, a full nodal dissection with para-aortic staging may be considered but treatment decisions will be left to the individual treating surgeons. Additionally, completion radical hysterectomy, radicality, fertility-sparing surgery will be performed according to the tumor and patient's characteristics. However, operative and post-operative treatment should comply with national or international guidelines. Open and minimally invasive access are allowed to accomplish the hysterectomy. Teams using minimally invasive surgery should be properly trained (according to data provided by the center). Results of the recently published locally advanced cervical cancer randomized trial should be discussed with patients.12 Patients will be stratified according to use of open minimally invasive surgery access.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Memorial Sloan Kettering Hospital algorithm.4 Surgical algorithm for early cervical cancer. ^aIntracervical injection with isosulfan blue dye, 99m-technetium, or both; ^bincluding interiliac/sub-aortic nodes; ^cexceptions made for select cases, see text. H&E, hematoxylin and eosin staining; LND, lymphadenectomy; SLN, sentinel lymph node.

SLNs will undergo permanent pathologic ultrastaging with serial sectioning each 200 μm, and staining with hematoxylin-eosin-safran. SLNs negative with hematoxylin-eosin-safran, will undergo immunohistochemistry with anti-cytokeratin AE1-AE3. Non-SLNs will be sectioned once and examined after staining with hematoxylin-eosin-safran. The standard definition of isolated tumor cells, micrometastases, and macrometastases will be used.13

Patients will be followed up post-operatively at 30 days and then every 3 months during the first year, every 4 months during the second year, and then every 6 months to complete 5 years of follow -up. These visits will focus on morbidity related to the treatment, especially lower-limb and lymphatic morbidity, survival and health-related quality of life (EORTC QLQ-C30). No systematic imaging is required for screening of recurrence.

This trial has been funded by a grant of Institut National du Cancer (INCA)(Le programme hospitalier de recherche clinique Cancer) for the French group and for the international coordination. Each participating group will find its own funding.

Setting

This study will take place in France with the collaboration of the INCA-labeled French Cooperative Group ARCAGY-GINECO. This large randomized study will also take place widely in Europe through the ENGOT network (European Network of Gynecological Oncology Trial groups) and in non-European countries thanks to GCIG and CCRN (Cervix Cancer Research Network) accredited centers. Three hundred centers will participate.

We will include patients with early cervical cancer of squamous, adenosquamous, or adenocarcinoma histology. Stages Ia1 with lymphovascular space invasion to stage IIa1 will be included. The tumor size must be <40 mm on clinical examination and MRI. No suspicious node should be seen on MRI (RECIST 1.1). Patients must be ≥18 years of age, have an Eastern cooperative oncology group performance status of 0 to 2, have given written informed consent, and agree to comply with follow-up. For French patients, being affiliated to, or a beneficiary of, a social security category is mandatory. Exclusion criteria include pregnancy, previous pelvic or abdominal cancer, and previous chemotherapy or external beam radiation therapy for the cervical cancer (brachytherapy is allowed). Patients with a proven allergy to blue dye, isotope, or indocyanine green will not be included.

Sample Size

We aim to demonstrate the non-inferiority of SLN biopsy alone versus SLN biopsy + pelvic lymph node dissection for disease-free survival using a gatekeeping procedure to control the type I error rate. Using a 5% non-inferiority margin and a standard 3-year disease-free survival of 85%, 900 randomized patients are required to observe the required (unilateral α error of 5%, a statistical power of 80%, and a 5-year follow-up). An interim analysis will be planned, when at least 132 events will be observed to reject H0 or H1 using the O'Brien-Fleming and α spending function. Stopping guidelines for efficacy will be −2.532 on the Z scale and −0.346 for futility. For health-related quality of life, three dimensions of the European Organization for Research and Treatment QLQ-C30 and QLQ-CX24 will be targeted: global health status, pain, and physical functioning. At 3 years, to demonstrate a superiority of at least one of the three targeted dimensions without significant deterioration in at least one with a minimal clinically important difference in mean score of at least five points (SD of 20), and a bilateral α type 1 error of 0.016 (Bonferroni adjustment for three dimensions), 815 patients would be required with available health-related quality of life scores at 3 years to reach an 85% statistical power. Taking into account a 5% loss to follow-up, 950 patients are needed. Accrual will last for 3 years.14

Randomization and Blinding

Patients will be randomized in a 1:1 ratio using an interactive web response system in a 1:1 ratio of the two treatment arms: SLN biopsy alone versus SLN biopsy + pelvic lymph node dissection by minimization technique, and stratified by center and stage of the disease using the 2018 Federation Internationale de Gynécologie Obstétrique classification. A double-blind study is not feasible since the surgeon and treating physician will be aware of the details of the operation. Thus, the treatment will be blinded to the patients only.

Statistical Methods

Efficacy analyses will be conducted on an intention-to-treat population including all randomized patients whatever treatment received and whatever eligibility criteria. For non-inferiority analysis of disease-free survival, results will be also reported on a per protocol analysis. For the health-related quality of life analysis, a modified intention-to-treat population will be considered, including all randomized patients with at least the baseline health-related quality of life score available. Safety analyses will be conducted on all randomized patients, with at least one post-baseline safety assessment. This population will be considered for safety data and treatment exposure data.

All tests will be performed two-sided at a significance level of 5% (95% CI), with the exception of the tests for the primary endpoint of non-inferiority which will be one-sided at a statistical significance level of 5% (90% CI). Indeed, for health-related quality of life analyses, all tests will be performed two-sided at a significance level of 1.6%. For disease-free survival, the nominal significance levels for the interim and final analyses will be derived from the α spending function with O’Brien-Fleming boundaries, which are dependent on the information fraction in the intention-to-treat population.

Confidence intervals will be calculated. These will be two-sided with a confidence level of 95% with the exception of the estimates for the primary endpoint using a confidence level of 100%— - nominal α and quality of life using a CI of 100–1.6, that is, 98.4%.

Clinical and demographic data at baseline will be described by treatment arm using rules form. The statistical parameters mean, median, SD, IQR, and range will be presented for continuous baseline variables. For categorical baseline variables, number and percentages will be calculated.

The Kaplan-Meier method will be used to estimate disease-free survival and other time-to-event endpoints. For the primary objective of non-inferiority in disease-free survival, the 2-year disease-free survival rate will be computed with its 90% CI. Follow-up will be estimated using the reverse Kaplan-Meier method, and will be described using the median with its 95% CI. For the primary health-related quality of life analysis, a t-test will be used to compare the health-related quality of life level at 3 years according to treatment arm for each targeting dimension (global health status, pain, and physical functioning). The statistical level for significance will be 0.016. The normal distribution of the scores will be checked. For non-normal distributions, a non-parametric Mann-Whitney test will be used. Overall survival will be estimated using the Kaplan-Meier method, and will be described using the median with its 95% CI. A univariate Cox proportional hazards model will be used to estimate hazard ratios with 95% CI. Multivariate Cox analyses will be done in respect of the Peduzzi rule of one variable for 10 events. A univariate selection procedure will serve to identify eligible explanatory variables with a univariate Cox p value according to the Wald test of <0.10 as a potential prognostic value. For each Cox analysis, the proportionality assumption of the risk will be graphically checked.

For safety analyses, the report will take into account all adverse events observed during and after the acts performed or methods used. Categorical data will be summarized in contingency tables displaying frequencies and percentages. Continuous data will be presented using median, minimum, and maximum values. The safety data of the different strategies will be compared using the Kruskal-Wallis test or Fisher exact test or Χ² tests. Particular interest will be given to rates of grade 3–4 toxicities according to the NCI-CTCAE v4.