Rapid Access to Contrast-Enhanced spectral mammogRaphy in women recalled from breast cancer screening: the RACER trial study design

All women recalled for a suspicious breast lesion from our national screening program are eligible for inclusion. Inclusion criteria are the ability to provide written informed consent and being recalled from breast cancer screening during the 18-month study inclusion period. Excluded are women with a known allergy to iodine-based contrast agents and women at risk for developing contrast-induced nephropathy or women with known renal insufficiency, according to the current guidelines.

The primary outcome will be the accuracy assessed by diagnostic performance parameters, such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), receiver operating characteristics (ROC) curve, and area under the curve (AUC). Radiologists will prospectively provide a single BI-RADS classification for each exam, where a BI-RADS score of 1–3 will be considered ”benign” and 4-5 ”malignant”. The final BI-RADS score (BI-RADS 1–5), based on imaging, will be compared with the BI-RADS score given by the screening program. The accuracy assessed by diagnostic performance parameters will be assessed after the next screening round after approximately 2 years.

Secondary outcomes will be quality of life (QoL), days until final diagnosis, cost-effectiveness, and experienced patient anxiety during a follow-up of 18 months. Three validated questionnaires will be presented at six different time points (at study inclusion; after 2 and 4 weeks; and after 6, 12, and 18 months). QoL will be assessed by the Dutch version of the EuroQol five-dimension, five-level (EQ-5D-5 L) questionnaire, including a visual analog scale (EQ-VAS). This questionnaire is a preference-based instrument used to value health states [11]. The Dutch tariff for the EQ-5D-5L questionnaire, as established by Versteegh et al., will be used to calculate utility scores per health state [12]. Then the quality-adjusted life years (QALYs) will be modeled based on these utility scores. Resource use related to diagnostics will be collected during the trial.

Health-related anxiety, both state and trait anxiety, will be measured by the Dutch version of the State Trait Anxiety Inventory (STAI-DY-1 and STAI-DY-2) questionnaires. Each STAI has 20 items and will be rated on a 4-point Likert scale, scoring 20–80 points per STAI. A higher score corresponds to a higher anxiety level [13].

For the QoL and STAI scores, inter- and intra-patient differences over time from baseline will be compared including those between the intervention and control group. The scores will also be compared between the women with a follow-up exam after 6 or 12 months and those without this follow-up. Figure 1 shows the outcomes defined by the five outcome elements described by Saldanha et al. [14].

To reduce patient effort, the questionnaires are offered digitally, or by telephone, if desired by the participant. Data from the digital questionnaires is automatically stored in the data management system.

All recalled women announced at one of the participating centers will be contacted for study participation. Written informed consent will be obtained and the first EQ-5D-5 L and STAI questionnaires regarding anxiety and current health state will be completed before any other study procedure is carried out. Participants are also asked for their consent to use their data in possible future studies. See Additional file 1 for the informed consent form in Dutch.

Women will be randomized to undergo standard of care (control arm) or CESM (intervention arm). Minimization with stratification factors will be applied in the computer-generated randomization screening and enrollment application software ALEA (version 3.0.2083.212r; ALEA Clinical, Abcoude, the Netherlands). Randomization will be stratified by the following: predominant reason for recall (mass, calcifications, asymmetry, architectural distortion), recall BI-RADS score (BI-RADS 0 versus BI-RADS 4/5), and study center. Enrollment in the four centers, including the master randomization file, will be overseen by the Study Coordinator. During the inclusion visit, written consent will be conducted and data entry will be done by one of the research team members in each center to execute randomization using ALEA. Should a problem occur in including a patient or randomization, the Study Coordinator must be reached so that he can provide assistance. After allocation, patient and radiologist will be informed about the outcome; hence, they are not blinded. However, the radiologist is blinded for the outcomes between the two groups, since the final diagnosis and pathology are not yet known. The outcome assessors and data analysts are blinded for the judgment of the next screening round by the screening radiologists, since accuracy also depends on the next screening round outcome. Each participant is assigned a trial identification number for anonymization, to analyze the data, and further to be used in publications. In each participant’s hospital patient file a statement will be noted about study participation and the assigned intervention or control group. On request, study participation can be ended during the follow-up phase.

Women undergoing standard of care will have their screening FFDM re-evaluated, adding (special) FFDM views, DBT, or US if deemed necessary. In FFDM and DBT, the breast is compressed between a paddle and detector plate. The laterality of the breast and views per breast depend on the judgment of the radiologist. At least one additional view of the recalled side will be performed. Tissue sampling (CNB, VAB, FNAC) can be recommended to support or to invalidate suspicious findings on imaging. In case of inconclusive findings, the radiologist can consider follow-up in 6 and/or 12 months or single breast MRI. The use of CESM is not permitted in this study arm.

CESM is based on visualizing angiogenesis in tumor tissue using dual-energy mammography [15]. Prior to image acquisition, an intravenous catheter will be placed in the antecubital vein, after which its patency will be checked by a saline test bolus. A non-ionic, low-osmolar contrast agent consisting of either iopromide (Ultravist® 300, Bayer Healthcare, Berlin, Germany) or iobitridol (Xenetix® 300, Guerbet, Villepinte, France) will be administered at a dose of 1.5 mL/kg body weight. If an automatic injector is used instead of manual administration, the injection rate will be 2.5–3 mL/s, followed by a saline flush. Image acquisition is started after at least 2 min after contrast administration [16]. Although there is a limited risk of complications, such as hematoma or incorrect catheter placement, a previous retrospective study showed that the risks of adverse reactions to the contrast agents and contrast-induced nephropathy are negligible [17]. Two days after CESM, the patient file will be consulted to investigate whether any late (serious) adverse reactions to the contrast agent have occurred that need medical treatment.

Similar to FFDM, the breasts are positioned between a mammography paddle and detector plate, and the four standard views are made: a cranio-caudal (CC) and mediolateral oblique (MLO) view of each breast. Special views can be requested by the reviewing radiologist. A typical CESM exam consists (per breast exposure) of two images: a low- and a high-energy image, which are acquired within seconds [18]. These images are used in post-processing to acquire the recombined image, which shows areas of contrast accumulation (Fig. 2). The CESM image acquisition needs to be completed within 10–12 min after contrast administration [19]. As in the control group, US, tissue sampling, MRI, and follow-up can be considered in case of inconclusive findings.

For the control group, final diagnoses of the (recalled) breast lesions can be divided into four categories [3, 5]: (1) simple cysts; (2) superposition densities; (3) solid, benign masses; (4) (invasive) breast cancer or ductal carcinoma in situ (DCIS). The first three groups are defined as benign findings. To ascertain a final diagnosis of (1), targeted US is performed, followed by cyst aspiration and a second FFDM of that breast to confirm that the lesion has disappeared. Final diagnosis of (2) requires a minimum of one special view or DBT, including targeted US. To rule out false negative (FN) findings, 6 and/or 12 months follow-up or breast MRI can be considered. Final diagnosis for categories (3) and (4) is acquired with US-guided or stereotactic tissue sampling. If cancer or DCIS is diagnosed, the subject will have a ”true positive” (TP) finding. In subjects where no breast cancer or DCIS is diagnosed, the follow-up period of 2 years will determine the true disease status. If no interval cancers have been detected and the subject has not been recalled in the subsequent screening round, the case will be considered a ”true negative” (TN).

In the CESM group, diagnoses are acquired slightly differently. For (1), CESM will show an ”eclipse sign” which is pathognomonic for cysts [3]. No further action is needed. For category (2), a negative CESM exam with no suspect lesion on both low-energy and recombined images rules out (pre)malignant lesions due to CESM’s high negative predictive value [3, 5]. To acquire final diagnoses of (3) and (4), CESM and targeted US are performed including tissue sampling for pathological confirmation.

Since most benign recalls are caused by cysts or superposition densities, the investigated intervention will most likely result in fewer additional exams and tissue samplings among these recalls [3, 5]. Moreover, follow-up exams can be omitted, which is more patient-friendly, and hypothetically, more women will attend the subsequent screening round when they are examined with CESM.

A flow chart is presented in Fig. 3, summarizing the study design, the randomization process, and the establishment of the different diagnoses.

Prior research showed a specificity of FFDM of 40% in this population [5]. To enable detection of a clinically relevant increase of specificity by 15% (from 40% to 55%), 176 patients without malignant disease per group are required (power 80%, alpha = 5%). The prevalence of malignant lesions in women recalled from the breast screening program is about 30%; thus, 70% have benign lesions. Therefore, a total of 251 women per group (176/0.7) has to be included. To account for 5% loss to follow-up, the final number of patients to be included is 528 (502/0.95). The calculations were done with OpenEpi [20].

The primary objective of the study is to demonstrate that the use of CESM as a work-up tool for women recalled from breast cancer screening is more accurate compared to standard care consisting of conventional imaging. The secondary objectives are to evaluate whether this novel approach is a more patient-friendly and cost-effective strategy in the work-up of women recalled from breast cancer screening, requiring fewer days until final diagnosis and less additional imaging.

Radiologists will prospectively provide a single BI-RADS classification for each exam, where a BI-RADS score of 1–3 will be considered ”benign” and 4-5 ”malignant”. Based on this cutoff, the final BI-RADS score and recall BI-RADS score, the diagnostic performance parameters sensitivity, specificity, PPV, NPV, and AUC under the ROC curve will be assessed as the primary endpoint for both study arms. Differences in proportions between the two randomized groups will be tested for significance with a chi-square test, and the difference in AUCs will be tested using the method proposed by Hanley et al. [21].

Secondary outcomes are QoL, patient anxiety, and cost-effectiveness. The course of scores on QoL with QALY and patient anxiety (STAI) from baseline over time at the six time points will be visualized and differences between groups will be tested using mixed linear models, which account for correlations between repeated measurements. Single and multiple imputation will be used to replace missing data.

Decision-analytic modeling will be applied to estimate the cost-effectiveness of CESM compared to conventional imaging modalities FFDM or DBT for recalled women from the Dutch national screening program. The economic evaluation will be performed from both a health care and a societal perspective. Cost-effectiveness will be expressed as the incremental costs per QALY gained as outcome measure [22]. All resource use related to diagnostics will be registered. In addition, productivity loss will be measured with the iMTA Productivity Costs Questionnaire (iPCQ) at 4 weeks and at 6, 12, and 18 months [23]. Reference prices will be obtained from the Dutch manual for costing research, hospital financial department, or the literature [24]. Uncertainty surrounding the incremental costs per QALY will be analyzed with a non-parametric bootstrap analysis [25]. Results of the bootstrap analysis will be presented in cost-effectiveness planes and acceptability curves. A probabilistic sensitivity analysis will be performed to examine different parameter uncertainties.

In Fig. 4 an overview, according to Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT), is given of all study activities, including the iPCQ assessments.

Participants are currently enrolled via protocol version number 6.1 (April 23rd 2019). The study protocol was approved by the Medical Ethics Review Committee of Maastricht University Medical Center+ in January 2018. The first participants were randomized in April 2018. Enrollment is expected to be completed before January 1st, 2021.