Intraoperative Avidination for Radionuclide Therapy: A Prospective New Development to Accelerate Radiotherapy in Breast Cancer

Breast conservative surgery, with sentinel node biopsy and postoperative regional radiotherapy, represents the treatment of choice in patients with early breast cancer (1, 2). Unfortunately, the duration of a standard course of whole-breast external beam radiation therapy (EBRT), followed by a boost to the tumor bed, is usually 6 to 8 weeks. This can represent a logistics problem for many patients, particularly those who are far from a radiation treatment facility (3). Alternative modalities of accelerated partial breast irradiation, such as MammoSite, breast brachytherapy, three-dimensional Conformal External Beam Radiotherapy, interstitial brachytherapy, and single fraction intraoperative radiotherapy, are currently under clinical evaluation (4–8).

Based on our previous clinical experience in locoregional treatment of peritoneal carcinomatosis and recurrent high-grade glioma using the avidin-biotin pretargeting technique (9–11), we foresaw a potential application of ⁹⁰Y-DOTA-biotin radionuclide therapy in breast cancer. We developed the Intraoperative Avidination for Radionuclide Therapy (IART) that relies on the avidin-biotin binding system. In fact, the “avidination” of the anatomic area of the tumor with native avidin, directly injected by the surgeon into and around the tumor bed, provides a target for the radiolabeled biotin i.v. injected 1 day later.

To prove the feasibility of this new procedure, we first did a pilot study to optimize the chase of biotinylated albumin to bind circulating avidin, thus restricting the presence of free avidin in the tumor bed. Consequently, 10 patients with early breast cancer, candidates for breast conservative surgery, were enrolled in a phase I trial specifically designed to study dosimetry and biodistribution of ⁹⁰Y-DOTA-biotin in target and nontarget organs through simulation studies based on ¹¹¹In-DOTA-biotin. Tumor resection was followed by avidin injection in the tumor bed. On the following day, patients were given ¹¹¹In-DOTA-biotin i.v. after appropriate chase of biotinylated albumin. Biokinetic studies were based on measured radioactivity in timed blood samples, 48-h urine collection, and through scintigraphic images. Our preliminary results suggested 20 mg of biotinylated albumin as the optimal chase dose to remove circulating avidin. The ¹¹¹In-DOTA-biotin simulation studies to predict dosimetry with ⁹⁰Y-DOTA-biotin showed that the mean absorbed dose in nontarget organs was <0.15 Gy/GBq, with the exception of kidneys and urinary bladder, where the doses (± SD) were 1.20 ± 0.42 and 1.39 ± 0.10 Gy/GBq, respectively. Interestingly, such absorbed dose reached 5.5 Gy/GBq (50% ISOROI) and 4.8 Gy/GBq (30% ISOROI) in the target area. Images showed early and long-lasting radioactive biotin uptake in the operated breast. Rapid blood clearance and urine excretion were observed. Such findings led to apply IART with ⁹⁰Y-DOTA-biotin for the first time in a patient with local breast cancer recurrence, showing an objective response to this new targeted locoregional therapy.

Native avidin (10 mg/mL), biotinylated human serum albumin (HSA-biot; 20 mg/mL), and DOTA-biotin ligand (2 mg/mL, product code ST2210) dissolved in saline were supplied by Sigma-Tau S.p.A. i.f.r. Anhydrous sodium acetate (NaAc), acetic acid, and diethylenetriaminepentaacetic acid were purchased from Sigma-Aldrich Italy at the highest available purity. A NaAc buffer at 1 mol/L (pH 5.0) was prepared using MilliQ water (ρ = 18 MΩ/cm) as previously described (12). Indium-111 chloride (¹¹¹InCl₃) in 0.05 N HCl was purchased from Mallinckrodt Medical B.V., whereas yttrium-90 chloride (⁹⁰YCl₃) in 0.04N HCl was obtained from Perkin-Elmer, Inc.

The radiolabeling of DOTA-biotin was carried out following a previously described procedure (12). Briefly, a volume of NaAc buffer, equal to that of the radionuclide MCl₃ (M = ¹¹¹In or ⁹⁰Y) solution, was mixed with a quantity of DOTA-biotin (Mw = 715) to achieve a specific activity of 2.65 MBq/nmol. The buffered DOTA-biotin was added to the MCl₃ vial and then incubated in a water bath at 95°C for 30 min. After incubation, radiochemical purity was assessed by Silica Gel Instant Thin Layer Chromatography (ITLC-SG, Gelman); an aliquot of the radiolabeled solution was mixed with a molar excess of avidin and diethylenetriaminepentaacetic acid, then spotted in triplicate. Quantitation of ¹¹¹In/⁹⁰Y–radiolabeled DOTA-biotin (R_f = 0) and free amount of ¹¹¹In/⁹⁰Y complexed to diethylenetriaminepentaacetic acid (R_f = 1) was carried out by high-performance storage phosphor screen (Cyclone, Packard BioScience).

Fifteen patients with breast cancer at stage T_1-2N_0-1, suitable for a quadrantectomy and axillary dissection (or sentinel lymph node biopsy), gave signed, written informed consent to participate in our study. In particular, 5 patients were randomly selected for a pilot study for B-HSA chase optimization and 10 were enrolled in a phase I trial. Inclusion criteria were performance status (Eastern Cooperative Oncology Group), 0-2; age, 18 to 75 years; life expectancy, >1 year; and general clinical conditions adequate for comprehension of informed consent and study compliance.

Exclusion criteria were performance status (Eastern Cooperative Oncology Group) >2, previous biopsy or surgically removed tumor, severe psychiatric disorders, pregnancy, and breast-feeding.

The study was approved by the Local Ethics Committee (IEO S208/504). The patient undergoing the pilot therapeutic study had advanced breast cancer with bone and lung metastases and local cancer recurrence, characterized by little lumps under the skin of the right thorax and reddish-pink rashes and swelling of the left breast area.

To correctly set up the phase I trial, we tested the chase while varying the quantity and volume of administered avidin and the gap time between avidin and biotin injection. In particular, avidin doses ranged from 50 to 150 mg diluted in 20 to 40 mL. One patient was not given biotinylated albumin (HSA-biot), whereas the others received a dose ranging from 15 to 25 mg, 3 to 10 min before ¹¹¹In-DOTA-biotin administration. Liver and breast uptake, expressed as percentage of injected radioactivity, were monitored.

Intrasurgical phase. In the phase I trial, the surgeon administered 100 mg native avidin with three 10 mL syringes as 1 cm-apart multiple injections into the tumoral-peritumoral bed and periareolar site, where necessary, to percolate the index quadrant. Avidin injection occurred after either sentinel node biopsy or axillary dissection.

In the pilot therapeutic study, the dose for the subdermal avidination of the skin lesion was 150 mg.

Postsurgical phase. Sixteen to 24 h after surgery, patients were given radiolabeled biotin i.v. as a slow bolus injection. ¹¹¹In-DOTA-biotin activity was 108 ± 9 MBq for patients (n = 10) enrolled in the phase I trial.

To chase the excess of free avidin, which had been drained overnight by the lymphatic system toward the blood stream, patients were given 20 mg/5 mL of HSA-biot via i.v. injection 10 min before radioactive biotin administration. In the pilot therapeutic study, the patient received 25 mg of HSA-biot as chase and, 10 min later, she was co-injected i.v. with 1.2 GBq ⁹⁰Y-DOTA-biotin for therapeutic purposes and 130 MBq ¹¹¹In-DOTA-biotin for biodistribution purposes.

To determine ¹¹¹In-DOTA-biotin blood clearance and excretion rate, blood samples were withdrawn at t = 5, 10, 30, and 60 min and t = 3, 5, 12, 16, 24, 36, and 48 h, assuming t = 0 at injection time. Urine was collected over 48 h. The urinary tract was assumed to be the only elimination route (13).

Radioactivity in blood and urine samples was measured with a gamma counter (Cobra II, Perkin-Elmer) with a 20% reading window centered on the 173-keV peak of ¹¹¹In. The efficiency of the gamma counter was determined by counting a calibrated source of ¹¹¹In with the same geometry used for biological samples. The radioactivity measurements were corrected for physical decay at injection time and expressed as a percentage of injected activity (IA%) versus time. The time-activity curve in the blood was fitted with three-exponential function as described (13). The activity in urine samples was used to obtain the cumulative IA% versus time eliminated through the kidneys.

Whole-body transmission scans with ⁵⁷Co-flood source were done for attenuation correction purposes. At 1, 4, 24, 36, and 48 h after a slow bolus injection of ¹¹¹In-DOTA-biotin, standard scintigraphic whole-body scans were acquired in anterior and posterior views with a double-head gamma camera (GE Millennium VG) equipped with a medium energy general purpose collimator.

Whole-body images were analyzed by the conjugated view method (14). Regions of interest were drawn manually over the whole body, breast region, and source organs (kidneys, heart, and lungs). The same set of regions of interest was used for all scans. Counts from the gamma-camera images were corrected for background, scatter, and physical decay; biological time-activity curves were obtained.

To further detect ¹¹¹In-DOTA-biotin uptake and distribution in the mammal gland and surrounding tissues, a single-photon emission computed tomography scan was acquired over the breast region at t = 16 h.

The time-activity curve in the breast gland was created from single-photon emission computed tomography and whole-body images. The breast region was divided into three distinct areas: the high-uptake area (i.e., 50% ISOROI, where all pixels/voxels have more than 50% of maximum counts); the mean-uptake area (i.e., area between 50% and 30% ISOROI); and the low-uptake area (i.e., area between 30% and 10% ISOROI).

SAAM II software (15) was used to depict the kinetics of the radioconjugate by an appropriate compartmental model and to obtain the number of disintegrations [ND(h)] occurring in the source organs for ⁹⁰Y, on the assumption that the kinetics of biotin with either ¹¹¹In or ⁹⁰Y would be identical.

Dose calculations were done for each patient according to the medical internal radiation dose formalism using OLINDA/EXM software (16), entering the ND(h) for all source organs. The absorbed doses were calculated according to the patient/phantom mass ratio.

The self-dose in the breast area was calculated assuming uniform activity distribution in spherical shaped tumors (Spheres Model available in OLINDA/EXM). The ND(h) for the tumor area was derived from the breast area time-activity curves.

The ND(h) in the red marrow was calculated from the ND(h) in blood, assuming that no specific uptake would occur in bone marrow. Uniform activity distribution as well as equivalent clearance in blood and red marrow was also assumed. Due to the small size of the radiolabel, specific activity in red marrow was considered equal to the specific activity in blood (17).

To calculate the absorbed dose to the bladder wall, the ND(h) for bladder contents was calculated according to the dynamic urinary bladder model (18), based on the experimental curve of the cumulative activity eliminated in the urine. The bladder was assumed to void every 1.5 h in the first 8 h after injection (forced diuresis) and at every 4.8-h interval thereafter.

The linear-quadratic model enabled a comparison of doses released through EBRT to those from IART and analysis of possible effects on nontarget organs (19, 20). In fact, we adopted the biological effective dose (BED) expression defined by:

where D_i is the dose delivered per cycle i; T_{1/2 rep} is the repair half-time of sublethal damage; and T_{1/2 eff} is the effective half-life of the radiopharmaceutical in the specific tissue. The α/β ratio relates the intrinsic radiosensitivity (α) and the potential sparing capacity (β) to a specified tissue or effect.

For the breast region, we used T_{1/2 rep} = 1.5 h, α/β = 10 Gy (21); for the kidneys, we used T_{1/2 rep} = 2.8 h, α/β = 2.6 Gy (22).

DOTA-biotin was successfully labeled with ¹¹¹In/⁹⁰Y, routinely achieving radiochemical purities ≥99.0%.

Patients ranged in age from 33 to 64 years. All had good performance status (Eastern Cooperative Oncology Group score = 0). Postoperative staging according to tumor-node-metastasis classification showed six T_1c (1-2 cm), one T_1mic (<0.1 cm), two T_is, and two T_1b.

Breast and liver uptake of ¹¹¹In-DOTA-biotin was calculated as the percentage of injected activity to attain the best combination by associating avidin concentration, HSA-biot mass, and injection time schedule. Table 1 summarizes the variables for each of the five patients. The absence of HSA-biot (case no. 1) led to rather high (>9%) liver uptake (Fig. 1A). Similarly, a short gap time between HSA-biot and ¹¹¹In-DOTA-biotin injections led to liver uptake (Fig. 1B; case no. 2: 5% and case no. 3: 2%). A low mass of avidin (50 mg) was associated with inadequate breast uptake (case no. 4: 0.4%) whereas 100-mg avidin seemed to be the optimal dose to reach satisfactory breast uptake with a 20-mg HSA-biot injection 10 min before radiolabeled biotin administration (case no. 5: >4%; Fig. 1C).

An avidin dose ratio of 5:1 was considered appropriate for a successful chase and led to 20 mg of HSA-biot.

Patients enrolled in the phase I trial showed no clinical adverse reactions: the intrasurgical injection of avidin was well tolerated and no side effects were observed after the i.v. injection of ¹¹¹In-DOTA-biotin.

Imaging and biodistribution. The scintigraphic images showed that the radiolabeled compound remained well localized within the operated breast area with very negligible background in other tissues (Fig. 2). The blood clearance was fast, the IA% in the blood at t = 24 h was 0.15 ± 0.10%, whereas the cumulative activity excreted in the urine was 81.4 ± 12.2% IA 24 h after injection.

Dosimetry and biological effective dose. Using SAAM II software, the ND(h) that occurred in the source organs for ⁹⁰Y was extrapolated (Table 2). The highest absorbed doses were in the urinary bladder wall (1.4 ± 0.1 Gy/GBq) and kidneys (1.2 ± 0.4 Gy/GBq). The estimated absorbed dose to the red marrow was 0.05 ± 0.02 Gy/GBq.

In seven patients, the maximum uptake in breast area was at t = 4 h. The other three had maximum uptake within 24 h after injection. The breast area time-activity curve revealed an early and stable radioactive uptake in the target area (Fig. 3). The mean value of the ND(h) for the mean-uptake region of the target tissue was 1.77 h, ranging from 1.27 to 2.41 h; the mean value of the absorbed dose to the mean-uptake breast region was 4.8 Gy/GBq, ranging from 3.3 to 6.2 Gy/GBq.

For the breast region, the mean T_{1/2 eff} equaled 47.3 ± 6.7 h, and therefore, the biological effective dose was 4.8 ± 1.1 Gy/GBq; for the kidneys, the mean T_{1/2 eff} was 8.6 ± 4.1 h with biological effective dose at 1.3 ± 0.5 Gy/GBq.

From the data of the pilot therapeutic case treated by i.v. co-injection of 1.2 GBq of ⁹⁰Y-DOTA-biotin and 130 MBq of ¹¹¹In-DOTA-biotin (Fig. 4A), we extrapolated that the absorbed dose was 1.8 Gy in kidneys and 2.5 Gy in urinary bladder wall. The target area received a dose ranging from 8 to 10 Gy. The patient had an objective local response to IART therapy (Fig. 4B), documented by clinical inspection lasting more than a year. Moreover, the patient had complete pain remission in the chest with consequent withdrawal of pain relief drugs (morphine).

In an attempt to provide quality of life improvement through investigating patient-customized anticancer therapies, we propose IART as a new procedure for partial breast irradiation in candidates for conservative breast surgery.

The simple two-step method of intraoperative locoregional avidin injection (first step) and ⁹⁰Y-DOTA-biotin i.v. administration after 16 to 24 h (second step) relies on avidin-biotin chemical affinity and Yttrium radioisotope properties. In fact, avidin is a 66-kDa glycosylated and positively charged (isoelectric point ≅ 10) glycoprotein with extremely high affinity for the 244-Da vitamin H, biotin (k_d = 10⁻¹⁵ mol/L; ref. 23). Surgical intervention causes an inflammation reaction that increases capillary permeability and triggers cation-exchanging flows (24). Therefore, at the injection site, avidin is retained for several days and may act as “artificial receptor for radiolabeled biotin” (25, 26). Preclinical animal studies and prior clinical experience show that avidin injection in surgically created cavities is well tolerated (27–30).

On the other hand, biotin can be easily labeled with ⁹⁰Y or ¹⁷⁷Lu through the chelating agent N,N′,N″,N‴-1,4,7,10-tetraazacyclododecan-1,4,7,10-tetraacetic acid (DOTA). ⁹⁰Y is a high-energy pure β-emitter (E_ave, β = 0.935 MeV), with a relatively short physical half-life (T_1/2 = 64.1 h) and a long penetration range in tissue (R_max = 11.3 mm). Thus, ⁹⁰Y-DOTA-biotin is particularly suitable for radionuclide therapy, with the likelihood of killing the majority of neoplastic cells related to the so-called cross-fire effect (31), whereas ¹⁷⁷Lu may be particularly suitable for microscopic residual disease or nipple sparing mastectomy, where it is crucial to avoid a high radiation dose to the derma. Avidin-biotin–based pretargeting methods have been used in clinical imaging and experimental therapy trials for more than a decade (11, 30).

This study suggests that the IART procedure succeeds in placing interstitial receptors (tissue avidination) able to bind labeled DOTA-biotin 1 day after breast conservative surgery.

The scintigraphic images showed a fast and long-lasting uptake of labeled DOTA-biotin at the operated breast site. This area is therefore suitable for an efficient, targeted therapy. Through simulation studies based on in vivo pharmacokinetics and biodistribution of ¹¹¹In-DOTA-biotin, we estimated the absorbed doses and biological effective dose to the breast region and normal organs during treatment with ⁹⁰Y-DOTA-biotin.

In the phase I trial, pharmacokinetics of labeled DOTA-biotin showed a rapid renal elimination, which, however, resulted in a moderate dose to the kidney. The absorbed dose to the most involved nontarget organs (urinary bladder and kidneys) was far from the threshold doses of tissue side effects, according to the EBRT experience (32). However, these threshold doses are not suitable for internal radionuclide therapies because the radiation dose delivered during such therapy differs in many aspects from that delivered by external beam irradiation (33). Recently, the linear-quadratic model has been used to quantify the dose-rate sparing concepts expressly for radionuclide therapy, taking into account the different dose rate as compared with EBRT (34–36). In particular, for an injection of 3.7 GBq, the absorbed dose to the kidneys was 4.4 ± 1.5 Gy, whereas the biological effective dose was 4.8 ± 1.8 Gy, which is clearly lower than the threshold dose for the reported kidney toxicity (22, 37).

The pharmacokinetics herein reported is useful to calculate the optimal injected activity of ⁹⁰Y-DOTA-biotin. The radiation dose released to the breast region is ∼5 Gy/GBq (biological effective dose, 4.8 ± 1.1 Gy/GBq) suitable for targeted radionuclide therapy. The estimated biological effective dose, according to the linear-quadratic model, would facilitate integration with EBRT treatment planning. IART has the great advantage of potentially being applied to any breast cancer treatable with conservative surgery; multifocality, tumor location, and size are not limiting factors. The surgeon injects avidin all around the tumor bed, without margin limitations or constraints. IART can be applied in all institutions where breast surgery is done and a nuclear medicine unit is available.

In our opinion, the combined use of IART followed by reduced EBRT is a valid approach to an easier, patient-tailored accelerated irradiation after breast conservative surgery.

Based on the pilot therapeutic study results and dosimetry data herein reported, we consider further clinical development of randomized trials with IART in combination with EBRT fully justified.