Cost-Effectiveness of Radar Localisation Versus Wire Localisation for Wide Local Excision of Non-palpable Breast Cancer

This was a single-institution study of prospective patients with early breast cancer undergoing WLE using RL from February 2021 to April 2023. It was registered on the Australian New Zealand Clinical Trials Registry (ACTRN12624000068561) and followed the Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) guidelines.¹² Inclusion criteria were female patients ≥18 years of age with a non-palpable breast lesion(s) (≤2 cm on preoperative imaging) that underwent WLE. Lesions >2 cm in size were excluded because of treatment variability, including neoadjuvant chemotherapy, which could potentially alter tumour size for targeting and final specimen volume.³ Exclusion criteria were current pregnancy or lactation, previous breast or axillary surgery, previous neoadjuvant chemotherapy, or known allergy to components of SCOUT, such as nickel. Patients who met inclusion criteria underwent WLE by using SCOUT RL (Supplementary material). This cohort was compared to a retrospective cohort of consecutive patients, meeting the same inclusion criteria (except for allergy to components of SCOUT), who underwent WLE using standard WL.

Radar localisation involved the SCOUT Surgical Guidance System (Merit Medical Systems, Inc. South Jordan, UT), a radar reflector and an infrared light-emitting probe that guides the surgeon to the lesion. It consists of a 12-mm fiducial reflector preloaded into a 16-gauge introducer needle, and a console and handpiece system, which localises the reflector. The reflector contains an infrared light receptor, transistor switch, and two nitinol antennae. The handpiece generates an audible signal when positioned over the reflector. Preoperative localisation involved percutaneous insertion of the reflector into the breast after local anaesthesia and under ultrasound guidance by an interventional breast radiologist. The reflector was deployed adjacent to or within the targeted lesion or adjacent to the clip. Post-procedure mammogram confirmed satisfactory placement of the reflector (Fig. 1).^2,3,13‐16

Wire localisation involved the Kopans wire, a stainless steel, 0.3-mm diameter wire with a proximal reinforced portion and a spring hook to anchor to the tissue surrounding the lesion. Preoperative localisation involved a planning ultrasound to locate the lesion. After local anaesthesia, and under ultrasound guidance, the hookwire was advanced through the lesion and the needle deployed by an interventional breast radiologist. Four to 6 cm of the wire was left to protrude from the skin, and a mammogram was used to confirm the position following the procedure.¹⁷

Wide local excision was performed in the usual manner with a periareolar, inframammary, radial, or semicircular incision depending on the tumour location and wire entry point for WL cases. Dissection was performed with a combination of diathermy and scissors to excise the lump with an approximately 1-cm cuff of normal tissue to aim for clear margins. Dissection was guided by using audible cues from the SCOUT probe for the RL cases and by using the hookwire for the WL cases. The specimen was orientated, marked with sutures or clips, and examined macroscopically and with x-ray to confirm the lesion within the specimen. Local anaesthesia was given, and a layered closure was performed.

Specialist breast pathologists provided histopathology reports, including surgical pathology and margin status. “Positive margins” and “close margins” were defined as tumour on ink and tumour ≤2 mm from ink, respectively, for invasive and in situ pathology.

The study primary outcomes were margin status, re-excision rates, and surgery delays related to preoperative localisation. Theatre delay averted was the measure of effectiveness as it had significant implications on resource use. Other outcomes were tumour specimen weight and volume between RL and WL techniques. Demographic information included age, American Society of Anesthesiologists Classification (ASA) status, body mass index (BMI), and previous surgery or neoadjuvant therapy. The number of days before surgery that the reflector or hookwire was inserted, and any complication during and after insertion were recorded. Operative duration, length of hospital stay, and complications up to 90 days postoperative were documented. Characteristics of the tumour specimen included tumour type, grade, size, weight and volume, ER/PR/HER2/Ki67 status, histologic type, margin status, and re-excision rates.

All costs were in 2023 Australian dollars (AUD$, Supplemental Table). The perspective of a “third-party” payer (Medicare), also known as a “healthcare payer” perspective, was adopted. A break-even point analysis was performed by using the equation: “Number of cases to break-even” = “Fixed Cost”/ (“Savings per case” − “Variable cost”).^18,19 The fixed cost was the purchase price of the SCOUT Surgical Guidance System (AUD$77,150). The variable cost was the cost of each use of SCOUT, including the reflector and delivery unit (AUD$498), and for each use of WL, including the hookwire (AUD$61.50). The savings per case was calculated by using: “Savings per case” = “Average surgery delay cost” x “Surgery delay reduction rate.” The cost of surgery delay because of preoperative localisation scheduling was calculated based on average costs of time (minutes) wasted in theatre by using microcosting, including costs for personnel and consumables related to each technique. The surgery delay reduction rate was the difference in rate of surgery delay between the RL and WL cohorts.^18,20

Cost-utility analysis measures included quality-adjusted life-years (QALYs), which is the metric for both quality and quantity of life. Quality of life is determined by a patient’s preference towards a given health outcome (“health state”).²¹ QALYs provide a common unit of measurement that allows comparison of the value and clinical effectiveness of surgical interventions, which each have their own utility scores. Utility measures the success or failure of a surgical intervention based on how well the intervention impacts a patient’s final health outcome.^18,22 A successful localisation for WLE and a surgery delay were defined as distinct “health states” with associated costs, probabilities, and utilities for use in the decision model. The costs for preoperative localisation and WLE were based on Medicare item numbers. Probabilities associated with health states were obtained for surgery with RL and with WL. Utilities used were obtained from surveys of surgeons.²³ These are ratings of their preference for a given health state on a scale of 0 to 1, where 0 equals death and 1 equals perfect life. The utilities were converted to QALYs for all health states: “QALY” = (“Utility of health state”) x (“Duration of health state”) + (“Utility of successful procedure”) x (“Remaining life years”), where “Remaining life years” = “Average life expectancy” − “Average age of patient,” and “Duration of health state” = “Estimated recovery time in weeks”/52 weeks.^18,21 The remaining life years value was calculated from the estimation that a female patient undergoing WLE is 59 years old and has a life expectancy of 85 years based on Australian Institute of Health and Welfare statistics.¹ For a successful operation, defined as a successful localisation for WLE resulting in clear margins, the duration of health state was 0 while the duration for a delay in surgery was 1 day (1/365 years).

A decision model was created for these data and incorporated the costs, QALYs and probabilities of the health states (Fig. 2). “Expected values” for costs and QALYs were calculated by multiplying the probability of a health state by its actual cost and QALYs, respectively. The incremental cost-utility ratio (ICUR) was calculated by using: “ICUR” = (“Expected cost of RL” − “Expected cost of WL”) / (“Expected QALY of RL” − “Expected QALY of WL”).²¹ This ratio represents the added cost to extend a patient’s life by 1 year of perfect health. An intervention is “cost-effective” if the ICUR is greater than 0 and less than the “willingness to pay” for an added year of perfect health, which was defined as AUD$50,000.^21,22,24 A one-way sensitivity analysis was performed to evaluate the robustness of the baseline decision analysis by observing the effect on the ICUR as the complication rate with use of WL was varied from 0 to 1.

Continuous variables were presented as means with standard deviation (SD) or as median with interquartile ranges (IQR) and dichotomous and categorical data as frequencies with percentages. The t-test was used to compare the mean for continuous variables between the two groups. A chi-squared test was used to compare the proportions of categorical variables between the two groups. Results with p ≤ 0.05 were considered statistically significant. Statistical analysis was performed with RStudio, v2023.09.2.

A total of 220 patients that underwent WLE for non-palpable breast lesions were included in this study. The mean age was 59.4 years (standard deviation [SD] 11.3) and most (72.7%) had invasive ductal carcinoma on final histopathology. A total of 110 patients had preoperative localisation using RL and 110 patients had it with WL. There were no significant differences in demographics and tumour characteristics of patients in the RL and WL cohorts (Table 1).

Placement of the reflector for RL was performed a median 3 days before the day of surgery (range 1–20 days), whereas patients who had WL underwent hookwire placement the day of surgery (range 0–1 days). No significant issues were identified during insertion and localisation of the RL cases. There were four cases encountered during the hookwire insertion, which all involved the hookwire dislodging immediately after insertion. This was identified on the post-procedure mammogram, and a second hookwire placed successfully in all cases.

Median specimen weights were similar between the RL and WL cohorts (RL 27 g [IQR 15.4–39.8] versus WL 30.7 g [IQR 19.3–53], p = 0.454, respectively). Median specimen volumes were similar between the RL and WL cohorts (RL 50.4 cc [IQR 30–79.5] versus WL 60 cc [IQR 34–108], p = 0.377, respectively). Positive margin, close margin, and reexcision rates for the RL cohort were 11.8%, 32.7%, and 14.5%, respectively. The WL cohort was not significantly different with regard to positive margin, close margin, and re-excision rates of 17.3%, 34.5%, and 21.8%, respectively (Table 2). In the RL cohort, there were two superficial surgical site infections, which were managed conservatively with antibiotics: two haematomas of which one required evacuation in theatre, and one pneumonia treated with antibiotics. In the WL cohort, there were two haematomas, one of which required evacuation in theatre and two seromas both managed conservatively.

The surgery delay rate in the RL cohort was significantly less compared with the WL cohort (0% vs. 10%, respectively; p = 0.029; Table 2). These delays were related to preoperative localisation and were logistical issues because of scheduling of radiology and surgery. The overall cost of RL utilisation over the 2-year study period was AUD$402,281 AUD, which included the cost of each RL use and the reflector. The cost of a single use of RL per case was AUD$3657.10. The average cost of a surgery delay in the WL cohort was AUD$2317.86 due to a mean delay of 79.1 min (SD 50.2). Routine use of RL reduced the surgery delay complication rate by 10% and prevented such delays in 11 cases resulting in a gross cost savings of AUD$25,496.46 (Table 3).

The break-even point analysis demonstrated that 290 cases using RL are needed to be performed routinely to recover costs of the medical device based on the average surgery delay cost of AUD$2317.86 and a surgery delay reduction rate of 10%. The greater the number of cases performed with RL, the greater the potential savings when this break-even point is overcome. Baseline cost-utility analysis showed that although the cost of WLE with RL was greater than that with WL by AUD$566.92, use of RL yielded a greater clinical benefit of 1.15 QALYs and the ICUR was only AUD$492.97 (Table 4). The one-way sensitivity analysis demonstrated that using RL was more cost-effective when the surgery delay complication rate with WL was ≥10% (using a maximum willingness to pay of AUD$50,000 per QALY).