Targeted alpha therapy with astatine-211-labeled anti-PSCA A11 minibody shows antitumor efficacy in prostate cancer xenografts and bone microtumors

Purpose Targeted alpha therapy (TAT) is a promising treatment for micrometastatic and minimal residual cancer. We evaluated systemic α-radioimmunotherapy (α-RIT) of metastatic castration-resistant prostate cancer (mCRPC) using the α-particle emitter 211At-labeled to the anti-PSCA A11 minibody. A11 is specific for prostate stem cell antigen (PSCA), a cell surface glycoprotein which is overexpressed in more than 90% of both localized prostate cancer and bone metastases. Methods PC3-PSCA cells were implanted subcutaneously (s.c.) and intratibially (i.t) in nude mice. Efficacy of α-RIT (two fractions—14-day interval) was studied on s.c. macrotumors (0, 1.5 and 1.9 MBq) and on i.t. microtumors (~100–200 μm; 0, 0.8 or 1.5 MBq) by tumor-volume measurements. The injected activities for therapies were estimated from separate biodistribution and myelotoxicity studies. Results Tumor targeting of 211At-A11 was efficient and the effect on s.c. macrotumors was strong and dose-dependent. At 6 weeks, the mean tumor volumes for the treated groups, compared with controls, were reduced by approximately 85%. The separate myelotoxicity study following one single fraction showed reduced white blood cells (WBC) for all treated groups on day 6 after treatment. For the 0.8 and 1.5 MBq, the WBC reductions were transient and followed by recovery at day 13. For 2.4 MBq, a clear toxicity was observed and the mice were sacrificed on day 7. In the long-term follow-up of the 0.8 and 1.5 MBq-groups, blood counts on day 252 were normal and no signs of radiotoxicity observed. Efficacy on i.t. microtumors was evaluated in two experiments. In experiment 1, the tumor-free fraction (TFF) was 95% for both treated groups and significantly different (p < 0.05) from the controls at a TFF of 66%). In experiment 2, the difference in TFF was smaller, 32% for the treated group versus 20% for the controls. However, the difference in microtumor volume in experiment 2 was highly significant, 0.010 ± 0.003 mm3 versus 3.79 ± 1.24 mm3 (treated versus controls, respectively), i.e., a 99.7% reduction (p < 0.001). The different outcome in experiment 1 and 2 is most likely due to differences in microtumor sizes at therapy, or higher tumor-take in experiment 2 (where more cells were implanted). Conclusion Evaluating fractionated α-RIT with 211At-labeled anti-PSCA A11 minibody, we found clear growth inhibition on both macrotumors and intratibial microtumors. For mice treated with multiple fractions, we also observed radiotoxicity manifested by progressive loss in body weight at 30 to 90 days after treatment. Our findings are conceptually promising for a systemic TAT of mCRPC and warrant further investigations of 211At-labeled PSCA-directed vectors. Such studies should include methods to improve the therapeutic window, e.g., by implementing a pretargeted regimen of α-RIT or by altering the size of the targeting vector.


Introduction
The high energy, short-range α-particles are suitable in therapies of minimal residual and small-size disease, including widespread small-sized metastatic disease in the bone. Compared with treatments using beta-emitters, targeted alpha therapy (TAT) could minimize surrounding tissue toxicity and increase the radiation energy specifically delivered to tumor cells. The α-particles' characteristics have been advantageously exploited for palliative treatment of metastatic castration-resistant prostate cancer (mCRPC) with 223 Ra-dichloride [1] which targets the cancer cells indirectly through accumulation in the bone. Utilizing direct targeting in TAT, e.g., by alpharadioimmunotherapy (α-RIT), could further leverage αparticle-based treatments of mCRPC through increased efficacy and broadened indication. Studies have shown promising results using targeting of the prostate-specific membrane antigen (PSMA) in combination with the αemitters 225 Ac [2][3][4][5][6], 213 Bi [7] and 211 At [8,9]. While the results from reported and ongoing anti-PSMA-studies have been very encouraging, these reports have also pointed out limitations; some mCRPC patients have a low PSMA-expression [10], a fraction of pathologic positive lesion sites can be PSMA-negative [11], i.e., there is heterogeneity in the occurrence and expression of PSMA. This underlines the need for other targets complementary to PSMA. Prostate stem cell antigen (PSCA) is a cellsurface antigen expressed in normal prostate and overexpressed in prostate cancer (PC) tissues. It is detected iñ 90% of primary prostate cancers and the expression increases with high Gleason score, advanced stage and bone metastasis [12], i.e., PSCA is a putative target for targeted therapy of mCRPC [13].
We studied efficacy of α-RIT on mCRPC in mice, mimicking microscopic bone metastases, using an 80 kDa anti-PSCA antibody fragment (A11 minibody) [14,15], labeled with the α-emitter 211 At. A minibody is expected to have a fast pharmacokinetics relative to larger or full-sized antibody formats and thus could be a therapeutically favorable match with the short-lived 211 At (half-life 7.2 h). Efficient targeting of the 211 At-A11 minibody on PSCA-expressing tumors was demonstrated.
Following biodistribution and blood toxicity studies, we conducted fractionated systemic TAT and observed strong antitumor efficacy on both subcutaneous (s.c.) macrotumors and intratibial (i.t.) microtumors. This indicates that directly targeted TAT by α-RIT is a promising regimen for treatment of mCRPC and that further studies of 211 At-labeled agents are warranted, including anti-PSCA vectors like the anti-PSCA A11 minibody.
To establish a PSCA-expressing sub-clone, PC3 cells were transfected with ScaI-linearized pcDNA3.1 comprising the full-length human PSCA gene [17]. Stably transfected clones were isolated, expanded and screened for PSCA-expression by RT-qPCR (Additional file 4: Figure S4a). High-expression clones were evaluated for changes in cell proliferation with CyQuant (Invitrogen). The final screening of PSCA-PC3 clones was made by a cell-binding assay using radiolabeled A11 minibody (Additional file 4: Figure S4b), as described below.

Astatine-211, radiolabeling and immunoreactivity
Astatine-211 was produced at the PET and Cyclotron Unit, Copenhagen University Hospital, Denmark. Dry distillation and radiolabeling were performed as described before [18,19]. Prior to 211 At-labelling, the A11 minibody was modified via conjugation with m-MeATE (N-succinimidyl 3(trimethylstannyl) benzoate. The resulting immunoconjugate was then labeled with 211 At. A dry astatine residue (50-100 MBq) was first activated by adding N-iodosuccinimide in methanol:1% acetic acid. Then the immunoconjugate (100 μg) was added and reacted with the astatine for 60 s. The radiolabeled antibody fraction was isolated by passage over a NAP-5 column (GE Healthcare) eluted with PBS. Radiochemical purity of 211 At-A11 was analyzed by methanol precipitation. Stability and fragmentation analysis was performed at 4 h after radiolabeling, using size exclusion liquid chromatography, FPLC (Superdex 200, GE Healthcare, Sweden). Immunoreactive fraction (IRF) was determined using a cell assay. Serial 1:2-dilutions of PC3-PSCA cell suspensions (0.15625-10 million cells/mL) were incubated with 211 At-A11 (5 ng) at 8°C for 3 h. After centrifugation and cell pellet washing, the IRF was calculated from the double-inverse plot of specific binding (B/T) versus cell concentration. Injection solutions were prepared by diluting 211 At-A11 in PBS.

Biodistribution and uptake in macrotumors
Male nude mice (BALB/c nu/nu, 8 weeks old; Charles River) were inoculated s.c. with PC3-PSCA tumor cells (2 million in 200 μL medium) on the flank. The mean tumor volume at the study start was~200 mm 3 . At tumor volumes > 1300 mm 3 or body weight loss > 20%, mice were taken out of the study and sacrificed. This study was approved by the Gothenburg Ethical Committee for Animal Research (Ethical permit: 283-2011), and all animals were maintained as regulated by the Swedish Animal Welfare Agency. For biodistribution, 20 mice were injected intravenously (i.v.) in the tail with 211 At-A11 (260 ± 20 kBq in 0.12 mL PBS). At 1, 5, 9, 23 and 42 h post-injection (hpi), 4 mice per time point were sacrificed and the organs dissected, weighed and measured for 211 At activity using a gamma counter (Wizard 1480, Perkin Elmer). An extra cohort of 6 mice was used to investigate the effect (at 5 and 9 hpi) of pre-treatment with sodium perchlorate (NaClO 4 ) to reduce uptake of free 211 At in certain organs. The NaClO 4 (Sigma Aldrich Sweden AB) was injected intraperitoneally (1.2 μmol/g in 0.1 mL PBS) 24 h and 1 h before injection of the 211 At-A11 minibody.
At 1 and 5 hpi, bone marrow samples from the femoral bones were taken to measure the uptake in the bone marrow (BM). The uptake of 211 At (percentage of injected activity per gram tissue, %IA/g) was corrected for radioactive decay to the injection time. The BM-toblood-ratio (BMBLR) was calculated by dividing %IA/g for BM by that of the blood. Since a secure dissection of the thyroid is prone to error, we instead dissect the corresponding part of the throat. Anatomically, this part contains the thyroid, larynx, a part of the trachea and the related connective tissues. For estimation of uptake and absorbed dose to the thyroid, we then assumed that all activity measured for the throat was contained in the thyroid (using a standard weight of 3 mg).

Dosimetry
The cumulated activity (total number of decays, Ã) for each organ was calculated from time-activity plots (activity per gram versus hpi) of the biodistribution data without NaClO 4 pre-treatment. Including contributions only from α-particles, the mean absorbed dose (D) was calculated as: where m denotes tissue mass, φ α the absorbed fraction (set to 1) and Δ α the mean energy per 211 At decay (1.09 × 10 −12 J). Finally, the absorbed dose for each tissue was calculated as the sum of all injections. For biodistribution with NaClO 4 pre-treatment, biodistribution data is presented for 5 hpi and 9 hpi. To approximate absorbed doses following pre-treatment with NaClO 4 , theoretical biodistribution data were calculated for time points 1, 23 and 42 hpi. This was done by using the quotient of the %IA/g-values received with pre-treatment divided by the %IA/g-value without pre-treatment. The quotient found at 5 hpi, for each organ respectively, was used to calculate a theoretical %IA/g-value for the 1 hpi time point, and the quotient for 9 hpi was used for the 23 hpi and 42 hpi time points.

Myelotoxicity after alpha-RIT
Considering the bone marrow as the primary doselimiting organ, a separate cohort of mice was used to estimate the maximum tolerable activity (MTA) in the therapy studies. The mean absorbed dose (Gy/MBq) to the BM was calculated from the %IA/g in the blood using the BMBLR at 5 and 9 hpi. Three groups (5 mice per group) were i.v.-injected with 0.8, 1.5 or 2.4 MBq of 211 At-A11 (corresponding to BM absorbed doses of 1, 2 and 3 Gy, respectively). The cell count for white blood cells (WBC), red blood cells (RBC) and platelets (PLT), as well as the hemoglobin (HGB), was measured before (day 1) and after treatment (day 6 and 13), by tail vein blood samples analyzed using a microcell counter (F-820; Sysmex, Kobe, Japan).

Alpha-RIT of macrotumors
For α-RIT of s.c. macrotumors, the animals were pretreated with NaClO 4 (twice before each treatment, described above) and then treated twice, 14 days apart, allowing for BM recovery in-between treatments. Three weeks after cell implantation (tumor volume~200 mm 3 ), the 211 At-A11 was i.v.-injected (1.5 or 1.9 MBq in 0.15 mL PBS, 10 mice per group). A control group (20 mice) received non-radiolabeled A11 minibody. Tumor growth and body weight was monitored weekly up to 120 days after therapy. The tumor volume (V) was determined from the largest (a) and the perpendicular diameter (b) and calculated as: V = (a × b 2 )/2.

Alpha-RIT of microtumors
Efficacy of α-RIT with 211 At-A11 was studied in two separate independent experiments. Bone microtumors were established by intratibial (i.t.) injections (right leg tibiae) with PC3-PSCA cells under anesthesia, as previously described [20]. After implantation, the animals received analgesics (Rimadyl; 5 mg/kg) for 5 days. Alpha-RIT of microtumors was conducted in two experiments (pretreatment with NaClO 4 as described above). In experiment 1, 63 mice were given 20,000 PC3-PSCA cells. Seven days later, the mice were divided into 3 groups and the treatment started. Group 1 (n = 19) and 2 (n = 20) were i.v.-injected with 211 At-A11 (0.8 MBq and 1.5 MBq, respectively, in 0.15 mL PBS). A control group (n = 24) was given non-radiolabeled A11 minibody. The same treatments were repeated 14 days later. Six weeks after the first treatment, the mice were sacrificed, the tibiae dissected, fixed in paraformaldehyde (3 days), decalcified in Parengy, serially sectioned (4-μm thickness), and stained with hematoxylin and eosin (Histocenter AB, Gothenburg, Sweden). Tumor growth was evaluated from the H&E slides in terms of presence or absence of tumor cells. In experiment 2, the injected number of cells was 5 times higher than in experiment 1, a total of 100,000 cells was given. At therapy, 7 days later, a cohort of 5 untreated mice was sacrificed to estimate the microtumor sizes at treatment. The treated group (n = 19) received two i.v. injections of 211 At-A11 (1.5 MBq in 0.15 mL PBS) 14 days apart. The control group (n = 20) received unlabeled A11 minibody. Six weeks after treatment, the tumor growth was determined by histological analysis, as described above. The tibiae were serially sectioned throughout, corresponding to~200 sections per tibia, depending on the sectioning angle. Each microtumor volume was estimated from the serial sections using multiple sections (4 μm to 100 μm apart, depending on tumor size). The tumors were approximated as an ellipse and the major, a, and minor, b, (perpendicular) semi-axes were measured. For each section, the tumor area, A, was calculated as A = a × b × π, and then the volume by multiplication with section thickness. Finally, the tumor volume was calculated by summing all sections encompassing the whole microtumor.

Statistical methods
Difference in organ uptake after biodistribution with and without NaClO4 pre-treatment was studied at two time points, 5 and 9 hpi, and analyzed (each time point respectively) using the unpaired Student's t test. For α-RIT on i.t. microtumors, the differences in tumor volume after treatment between independent groups were evaluated with Mann-Whitney U test. Differences in tumor frequencies between treated and control groups were analyzed with the Pearson's chi-squared test. P values < 0.05 were considered as statistically significant.

Biodistribution and uptake in macrotumors
Biodistribution data of 211 At-A11 is presented in Table 1. The uptake in PC3-PSCA macrotumors was fast, maximizing at 7.2%IA/g after 5 h, and then decreasing slowly to 5.6 %ID/g at 23 hpi. Blood clearance was relatively fast, as expected for a minibody. The tumor-to-blood ratio increased with time ( Fig. 1), reaching a ratio of 5 at 23 hpi and close to 10 at 42 hpi. Included in the plot for comparison is the tumor-to-blood ratio of 211 At-MX35-F(ab') 2 , for which complete eradication of macrotumors was found by fractionated systemic α-RIT in an ovarian cancer model [21]. Alpha camera imaging [22] of the intratumoral activity distribution at 3 hpi showed that 211 At-A11 was well distributed throughout the macrotumors (Fig. 2), with hot-spots corresponding to the vasculature. Pre-treatment with NaClO 4 decreased the uptake in most organs (Fig. 3), and was therefore given in the therapies. For stomach, salivary glands, throat, lungs and spleen, the decrease (mean of 5 hpi and 9 hpi) was 74%, 82%, 55%, 24% and 31%, respectively. A statistical comparison between the groups with and without NaClO 4 -pre-treatment is included in Table 1. It can be noted that for salivary glands, throat and stomach, NaClO4-pretreatment gave a significant decrease in uptake at both 5 hpi and 9 hpi. For the liver, spleen and intestines, there was a significant decrease at 5 hpi, but not at 9 hpi. For the blood, there was no difference at 5 hpi, but at 9 hpi, there was a significant increase. For tumors, finally, no significant differences were observed, neither at 5 hpi nor 9 hpi.
At day 13, the WBC for the 0.8 and 1.5 MBq group (6.6 ± 0.5 and 6.2 ± 0.8, respectively) had recovered to a level similar to the controls (7.1 ± 1.3). The effect on PLT was similar to that of the WBC, with clear reductions at day 6 for the 2.4 and 1.5 MBq groups (Additional file 3: Figure S3). No effect was seen on RBC or HGB. For the 2.4 MBq group, several mice showed abnormal signs (dehydration and malnutrition) at day 7 and were sacrificed. All mice in the 0.8 MBq and 1.5 MBq groups were kept for long-term study of myelotoxicity. At day 252, no sign of radiotoxicity was seen and both groups had normal WBC counts (10.7 ± 3.1 and 9.8 ± 1.2, respectively). The values for PLT, RBC and HGB were also normal. The MTA of 211 At-A11 minibody was estimated to be within 1.5-2.4 MBq.

Dosimetry
Absorbed doses to organs and macrotumors were calculated from biodistribution data with and without NaClO 4 pre-treatment and are shown in Table 2. Without NaClO 4 pre-treatment, the highest doses were found for the thyroid, stomach, salivary gland, followed by the lungs, spleen and the bladder. The dose to blood was 4.7 Gy/MBq, corresponding to a BM dose of 1.3 Gy/ MBq. The dose to macrotumors was 2.2 Gy/MBq. When pre-treatment with NaClO 4 was given, the estimated absorbed doses decreased for most organs except the blood and bone marrow, for which a very small increase was observed, by 2% and 8%, respectively. For the Uptake data (%IA/g ± SD) at different times after i.v. injection of 211 At-A11 minibody. At 5 and 9 hpi, data is presented with and without NaClO 4 pre-treatment.(n = 4) and difference between the groups, at 5 and 9 hpi respectively, analyzed using unpaired Student's t test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 %IA/g Percent of injected activity per g; i.v. Intraveneous; ns Not significantly different; N.D. Not determined Fig. 1 The tumor-to-blood ratio (of %IA/g) of 211 At-A11 on s.c. PC3-PSCA macrotumors plotted versus time after injection. Included for comparison is the tumor-to-blood ratio of 211 At-MX35-F(ab') 2 , which was shown to be therapeutically efficient on macrotumors in previous studies of ovarian cancer [21] salivary glands and stomach, the dose decreased by 82% and 75%, respectively and for the thyroid and spleen by 57% and 53%. A decrease in dose was found also for the bladder and lungs, by 36% and 27%, respectively. The dose to macrotumors decreased to 1.9 Gy/MBq, i.e., by 14% as did the prostate dose (37%).

Alpha-RIT on macrotumors
At treatment start, 20 days after implantation, the mean tumor volume was 240 mm 3 . Five weeks later (at day 55 after implantation), when 85% of the control mice were sacrificed due to exceeding tumor volume, the tumor volumes in both 1.5 MBq and 1.9 MBq groups (139 ± 41 mm 3 and 128 ± 34 mm 3 , mean ± SD, respectively) were significantly different from the controls (p < 0.001; Fig. 4a). At the study end, 90 days after implantation, the difference between the 1.5 MBq and 1.9 MBq groups was only moderate but significant (p < 0.05), 379 ± 82 mm 3 versus 143 ± 84 mm 3 , respectively. For all three groups, the age of the mice at the days mentioned above (20, 55 and 90) was 14, 19 and 24 weeks, respectively. At 70 days after treatment, both treated groups showed a clear weight loss (Fig. 4b).

Alpha-RIT on microtumors
In experiment 1, the treated groups received two i.v. fractions with 211 At-A11 (0.8 or 1.5 MBq mice,) starting at 7 days after cell implantation. Six weeks later, the number of mice with tumor-negative tibiae was 19 out of 20 and 18 out of 19, respectively, i.e., the tumor-free fraction (TFF) was 95% for both treated groups. For the controls (non-radiolabeled A11), the TFF was 66% (16 tumor-negative out of 24), i.e., also high, but statistically different from the treated (p < 0.05). The tumor volumes for the mice with tumors are shown in Fig. 5a. For the 8 control mice with tumors, the variation in tumor volume was large, with a mean volume of 0.89 ± 1.43 mm 3 . Only one single tumor per group was found for the treated mice, with volumes of 0.057 mm 3 and 0.0056 mm 3 (0.8 and 1.5 MBq, respectively). In experiment 2, starting at 7 days after implantation, two fractions with 211 At-A11 (1.5 MBq) were given and the results are shown in Fig. 5b. Five mice were sacrificed at treatment start and four of these (80%) had tumor-positive tibiae ( Fig. 5c and d), with a mean tumor volume of 0.018 ± 0.12 mm 3 , corresponding~100-400 μm diameter. At the study end, the TFF was 20% for the control group (non-radiolabeled A11) and 32% for the treated group, i.e., only a minor difference to the controls. However, the treatment had a strong effect on microtumor volume. The mean volume in the treated group was 0.015 ± 0.004 mm 3 , while for the untreated controls it was 4.74 ± 1.45 mm 3 , i.e., significantly larger (p < 0.001, Mann-Whitney U test) (Fig. 5c). This corresponds to a 99.7% decrease in microtumor volume. Examples of untreated and treated tumors at study end are shown in Fig. 5e and f, respectively.

Discussion
Prostate cancer (PC) is the most common cancer among men in the Western world and the second in cancerrelated death causes [23]. In the castration-resistant stage, current therapies prolong life with~2-6 months [24] and new treatments are needed. Metastases remain the major clinical problem in the treatment of mCRPC, being the main cause of pain and death. Due to the αparticle characteristics, TAT is well suited for the treatment of metastatic disease. In contrast to 223 Radichloride, a regimen with direct targeting of PC cells, e.g., by PSCA-targeted α-RIT, could be effective on all types of PC micrometastases, i.e., both sclerotic and lytic Fig. 2 The intratumoral activity distribution of 211 At-A11 in s.c. PSCA-PC3 macrotumors at 3 hpi (a) as studied by alpha camera imaging [22]. The color coded LUT is normalized so that 1.0 represent the mean activity for the whole tumor section. White scale bar indicate 1000 μm. Consecutive H&E-stained section (b). LUT look-up-table; H&E Hematoxylin and eosin bone lesions, distant metastases or remaining active cells in the prostate, as well on CTCs in the blood.
We investigated the efficacy of fractionated α-RIT with the anti-PSCA minibody A11, labeled with the α-emitter 211 At. We found strong growth inhibition on both s.c. macrotumors and on intratibial microtumors. In experiment 1 on microtumors, the fraction of tumor-free mice (TFF) 95% for both treated groups, but also the untreated group had a relatively high TFF of 66%. Due to the relatively moderate difference between treated and untreated mice in terms of TFF, any treatment effect of 211 At-A11 could be questioned. However, when we analyzed the efficacy in study 1 in terms of actual measured tumor volumes, we could observe a clear difference between the untreated controls versus both the treated groups (Fig. 5a). This difference was also statistically significant, both 0.8 MBq vs untreated and 1.5 MBq vs untreated. In experiment 2, the observed TFFs were markedly lower than in experiment 1, and the difference in terms of TFF was relatively small. But the difference in terms of mean tumor volume the difference was large and strongly significant, corresponding to a 99.7% decrease in microtumor volume for mice treated with 211 At-A11. The difference in the outcome of the two experiments is most likely due to differences in microtumor sizes at therapy, or higher tumor-take in experiment 2 (where more cells were implanted). In experiment 1, the tumor size at therapy was unknown, but in experiment 2 the microtumor diameters ranged from 100 μm to 400 μm. The distribution and diffusion of 211 At-A11 into the PC3-PSCA intratibial microtumors was not known. For microtumors with diameters > 150 μm (α- Fig. 3 The relative effect of pre-treatment with sodium perchlorate (2.5 mg NaClO 4 ·H 2 0) on the biodistribution of 211 At-A11 minibody. For each organ. the %IA/g-value received with pre-treatment was divided with the %IA/g-value without pre-treatment, respectively for the two time points Dose to bone marrow was estimated using a bone marrow-to-blood-ratio of 0.29 b Dose to thyroid was estimated from the throat (assuming that all activity was located in thyroid and using a standard weight of 3 mg) c Dose corresponding to pre-treatment with NaClO 4 , using biodistribution data measured at 5 hpi and 9 hpi, complemented with theoretically calculated data for 1, 23 and 42 hpi  (c and d), magnification 40×, scale bar: 500 μm (e and f)TFF tumor free fraction particle range~70 μm), the 211 At-A11 would need to be well distributed intratumorally if all cells should be irradiated and high TFF achieved. For pre-vascular microtumors, this is likely the case, but at the later stage, the intratumoral distribution can be negatively affected by heterogeneous vascularization [25]. For the s.c. macrotumors, the two treatments with 211 At-A11 induced a clear reduction in tumor volume that was significantly different from untreated mice. However, the strong inhibition of tumor growth was also accompanied by radiotoxicity manifested as a clear and progressive reduction in body weight. This indicates the administered activities for therapy of macrotumors were very close to being above maximum tolerable. Indeed, the group receiving 2.4 MBq in the pre-study on myelotoxicity showed acute radiotoxicity and had to be sacrificed at day 7 after injection. The biodistribution study showed a fast uptake in macrotumors of the 211 At-A11 minibody, maximizing at 5 hpi, and then decreased between 23 and 42 hpi. However, because of the 7.2 h halflife of 211 At, 90% of the absorbed tumor dose was achieved already at 23 hpi. Since the clearance from blood was fast, this indicates that the fast kinetics of the minibody format could be a better match with the shortlived 211 At, than would a full-sized IgG. Such comparison however, should also include smaller compounds, e.g., dia-, nano-or affibodies, i.e., could be addressed in future studies.
The observed growth inhibition of s.c. macrotumors indicated a favorable intratumoral activity distribution. Accordingly, alpha camera imaging [22] at 3 hpi revealed a more uniform distribution of the 211 At-A11 minibody than was previously observed in an ovarian cancer xenograft model for the similarly sized 211 At-MX35-F(ab) 2 [22]. We have previously shown that complete eradication of s.c. macrotumors could be achieved by fractionated α-RIT for mean absorbed tumor doses > 10 Gy [21]. By adding a third fraction to the current study, i.e., increasing the highest tumor dose to 10.8 Gy, complete eradication might have been reached.
For the microtumors, the absorbed doses were unknown. However, estimations can be made from three different assumptions; (i) since the microtumors were residing in the BM cavity the doses would be at least the same as for the BM (1.3 Gy/MBq); (ii) weighing in antibody-to-tumor binding, the microtumor doses would be at least equal to that of the macrotumors (2.2 Gy/ MBq), and (iii) since previous studies have shown that for the one and same cell line, the uptake in microtumors could be 20× higher than in macrotumors [26], i.e., a 20× higher absorbed dose for microtumors. Hence, for two injections of 1.5 MBq, these three scenarios would correspond to mean absorbed doses to the microtumors of 3.8 Gy, 6.6 Gy and 130 Gy, respectively. Tumor doses > 100 Gy have been observed for microtumors with a diameter of ≤ 100 μm [27].
For the normal organs, without NaClO 4 pre-treatment given, the highest absorbed doses were found for organs known to accumulate free 211 At, i.e., the thyroid, stomach, salivary gland and spleen. This could indicate catabolism of the labeled minibody or reduced in vivo stability of the 211 At-labeling. Internalization of the minibody could contribute to the occurrence of free 211 At, but PSCA is a slowly internalizing antigen [15]. As for the influence of radiolabeling, results from comparative biodistribution (Additional file 5: Fig. S5) showed that blood concentration was very similar for 211 At-labeled and 125 I-labeled (iodogen) A11 minibody, while the tumor uptake was lower for the 125 I-labeled version. Since 211 At-labeling of antibody fragments has been associated with reduced in vivo-stability we also compared the m-MeATE method to labeling with the boron cage B10 reagent (kindly provided by Dr. S. Wilbur, Seattle, US) [28], but for the latter, we observed an elevated uptake in the bone marrow at 1 hpi (Additional file 6: Fig. S6).
Pre-treatment with NaClO 4 markedly reduced the uptake in the 211 At-accumulating organs, decreasing the absorbed dose by 55-82%. For the blood and bone marrow, however, a small increase (2-8%) was observed. For tumors, the uptake at 5 hpi was decreased when NaClO 4 was given. While this decrease was not significant, it contributed to a 13% decrease in tumor dose. There was no gain in the therapeutic window (blood/bone marrow vs tumor) when NaClO 4 was given and the rationale for pre-treatment could be questioned. The gain will depend on which organ is anticipated to be dose-limiting, but most certainly, bone marrow is one of limiting organs. We have previously estimated the maximum tolerance dose (MTD) for the kidneys following systemic α-RIT with 211 At in mice [29] and reported a MTD of 10 Gy. In similar, we reported that the total absorbed dose needed for the eradication of macrotumors [21] was estimated to be 10 Gy, also. These numbers can be used as an example to compare the therapeutic window with or without NaClO 4 pre-treatment in the current study, i.e., anticipating the kidneys as the dose-limiting organ. Without pre-treatment, a total injected activity of 4.5 MBq would be required to reach a tumor dose of 10 Gy, which in turn would give the kidneys a dose of 9.5 Gy. With NaClO 4 pre-treatment given, the corresponding numbers would be 5.2 MBq injected and 8.9 Gy to the kidneys, i.e., a 7% gain in the therapeutic window. However, at this point, the organ tolerance doses for systemic α-irradiation are largely unknown as are the respective normal organ RBEs (relative biological effectiveness). We have observed (unpublished data) that apart from bone marrow and kidneys, also the lungs could be a dose-liming organ in TAT. Even though the whole-body payload from α-irradiation was significantly decreased by pre-treatment with NaClO 4 in our study, more studies are needed to evaluate the potential value of blocking agents. It should be noted that the absorbed doses discussed above were not corrected for any RBE of αradiation, but a factor of 5 is generally accepted.
Among the α-emitters of interest for human use, 211 At is a main candidate and we have explored it continuously for clinical use in α-RIT of ovarian cancer [30][31][32]. Astatine-211 offers a theranostic approach to TAT since the decay involves the emission of K X-rays (77-93 keV) allowing for in vivo γ-quantification by planar and SPECT imaging. Another advantage is that 211 At does not have a series of α-particle emitting daughters in its decay chain. This can be of importance for accurate risk analysis and dose-planning, especially for TAT regimens with curative intent.
Their rationale for directly cell-targeted TAT of mCRPC is strong as indicated by the promising results with anti-PSMA ligands labeled with α-emitters [2][3][4][5][6][7][8][9]. While PSMA is detected on the majority of prostate cancer cases, the expression can be heterogenic at all levels; patient, lesion [33] and circulating tumor cells (CTCs). One study reported that the fraction of patients having a PSMA-positive phenotype in PC relapses was 82.8% [34], i.e., 17% would be PSMA-negative. A similar fraction (16%) was found not eligible for 177 Lu-PSMA-617 therapy [10], due to low PSMA-expression or FDG discordance. An imaging study with 89 Zr-J591 reported that 36% of pathologic positive lesion sites were PSMAnegative [11]. Another study compared imaging with biomarker analysis of CTCs showing that while nearly 100% of the lesions (bone and soft tissue ) were PSMApositive, only 43% of the patients had PSMA-positive CTCs [35]. Further, it has been reported that for mCRPC patients who progress after conventional treatments, a low PSMA-expression (or discordant PET/CT) correlated with poor prognosis and short surviva l [36]. In a study using immunohistochemistry to analyze the expression of both PSMA and PSCA on patient lymph nodes and bone metastases, both antigens had high overall staining frequency (94-100%), and in some samples, PSCA had a higher score than PSMA [37]. All this underlines the need for other targets that can be complementary to, or even combined with, PSMA-targeting. In fact, dual targeting of two antigens, e.g., PSMA and PSCA, either as a cocktail or using a bispecific targeting agent, could potentiate future treatments of mCRPC further, as indicated by [38].

Conclusion
We evaluated the concept of systemic TAT for treatment of mCRPC. Using a fractionated regimen of α-RIT with 211 At-labeled anti-PSCA A11 minibody, we found strong growth inhibition on both macrotumors and intratibial microtumors. These findings are conceptually promising for systemic TAT of mCRPC and warrant further investigations of 211 At-labeled vectors, including anti-PSCA antibodies and other molecules having PSCA as the target for therapy. Such investigations should include further optimization of the therapeutic window, e.g., by implementing pre-targeting (PRIT) or by altering the size of the targeting vector.