Targeted α-particle–emitting radionuclides have great potential for the treatment of a broad range of cancers at different stages of progression. A platform that accurately measures cancer cellular sensitivity to α-particle irradiation could guide and accelerate clinical translation. Here, we performed high-content profiling of cellular survival following exposure to α-particles emitted from radium-223 (223Ra) using 28 genetically diverse human tumor cell lines. Significant variation in cellular sensitivity across tumor cells was observed. 223Ra was significantly more potent than sparsely ionizing irradiation, with a median relative biological effectiveness of 10.4 (IQR: 8.4–14.3). Cells that are the most resistant to γ radiation, such as Nrf2 gain-of-function mutant cells, were sensitive to α-particles. Combining these profiling results with genetic features, we identified several somatic copy-number alterations, gene mutations, and the basal expression of gene sets that correlated with radiation survival. Activating mutations in PIK3CA, a frequent event in cancer, decreased sensitivity to 223Ra. The identification of cellular and genetic determinants of sensitivity to 223Ra may guide the clinical incorporation of targeted α-particle emitters in the treatment of several cancer types.

Significance:

These findings address limitations in the preclinical guidance and prediction of radionuclide tumor sensitivity by identifying intrinsic cellular and genetic determinants of cancer cell survival following exposure to α-particle irradiation.

See related commentary by Sgouros, p. 5479

The linear accelerator, the most commonly used device in clinical radiotherapy practices, generates sparsely ionizing radiation in the form of X-rays or electrons (1). The penetrant but dispersed ionization tracks are maneuvered to conform to the shape of the target tumors using multiple incident beams that superpose to effect tumor cell death (2). However, ionization tracks rely on a confluence of variables to confer death. Tumor cells sustain lethal damage only when two or more lesions take place within one or two helical turns of a DNA strand (3–5). The frequency of clusters of damage also varies based on the orientation of the DNA, its compactness, and the probabilistic trajectories of the ionization tracks (6). Contributing to the uncertainty, the genetic variation across and within distinct cancer types has been shown to modulate the risk of tumor cellular death by either mitigating damage or facilitating its repair (7). The mere conditional ability of sparsely ionizing radiation to cause lethal DNA damage is reflected in the overall incomplete clinical local control across several cancer types (8).

Shortly after the discovery of the X-ray in 1895, Marie and Pierre Curie described the activity of radium and its more potent physiologic properties compared with X-rays (9). Short exposures of radium to the skin produced inflammation that exhibited similar effects to those obtained after much longer exposure to X-rays. This suggested that radium's emitted particles are substantially more potent than sparsely ionizing X-rays. It is now evident that short-range charged α-particles emitted by radium induce clustered DNA damage along their tracks, resulting in significantly more effective cell death per unit of absorbed dose (10–12). Despite the qualitatively observed potency, the exact relative biological effectiveness of α-particles compared with sparsely ionizing radiation across a panel of cancer cells remains unknown (13, 14), and it is unclear whether physical (e.g., cellular size, shape, and nuclear volume) or cancer genetic variables can modulate the survival of tumors to this more potent form of radiation.

The long half-life (∼1,600 years) of radium's most stable isotope, radium-226, precluded radium's medical use for over a century. However, an artificially generated isotope with a much shorter half-life, radium-223 (223Ra, half-life = 11.4 days), has been effectively incorporated into routine cancer therapeutic use. 223Ra, the first α-emitter approved by the FDA, takes advantage of radium's bone mimetic properties for the treatment of patients with castrate-resistant prostate cancer with bone metastases (15, 16). The established clinical efficacy of 223Ra has led to significant interest in expanding the use of α-particles to target cancers other than those that are localized in bone. The treatment of extraskeletal cancers can be achieved through the use of targeted radionuclide therapy, which use molecular carriers with high affinity to antigens on the surface of tumor cells (17–21). The ability to target α-emitting particles to visceral disease is poised to improve response rates across a range of cancers, including appreciably more radiation resistant solid tumors.

Despite the potential for the use of targeted α-emitting radionuclides in solid tumors, very little is known about the interplay between α-particles and tumor cell sensitivity. To date, there have not been extensive analyses of the cellular vulnerability to α-particle radiation within or across cancer types. Inhibition of DNA double-strand break repair has been shown to sensitize cancer cells to α-particle treatments (22, 23), suggesting that the composition of the cancer genome could regulate sensitivity. However, a genetic basis for response to α-particles across distinct tumors has yet to be established. An improved understanding of the relationship between the sensitivity of tumors to α-particles and their cellular and genetic characteristics can more appropriately inform their future application. These could include the tailoring of α-particle prescriptions and schedules, biomarker-guided patient selection, and/or using more precise drug/α-particle treatments.

Here, we develop an integrated imaging, microdosimetric, cellular, and genomic high-content platform that measures the survival across a diverse panel of tumor cells following exposure to α-particles and leverages cancer genetic data for biomarker identification.

Cell lines from the Cancer Cell Line Encyclopedia (CCLE) were authenticated per CCLE protocol (24) and grown in recommended media supplemented with 10% fetal bovine serum (Thermo Fisher Scientific) and 100 U/mL penicillin, 100 μg/mL of streptomycin, and 292 μg/mL L-glutamine (Corning). Immortalized bronchial epithelial BEAS-2B cells were purchased from ATCC and maintained in advanced DMEM/F12 media (Thermo Fisher Scientific) supplemented with 1% fetal bovine serum and 100 U/mL penicillin, 100 μg/mL of streptomycin, and 292 μg/mL L-glutamine. All cultures were maintained at 37°C in a humidified 5% CO2 atmosphere and tested to ensure absence of Mycoplasma. Plates were treated with γ-radiation delivered at 0.85 Gy/minute with a 137Cs source using a GammaCell 40 Exactor (Best Theratronics) or radium-223 dichloride (223RaCl2). The specific activity of 223Ra is 1.9 MBq/ng. The six-stage-decay of 223Ra to lead-207 (207Pb) occurs via short-lived daughters, and is accompanied by a number of α, β, and γ emissions with different energies and emission probabilities. The fraction of energy emitted from 223Ra and its daughters as α-particles is 95.3% (energy range, 5.0–7.5 MeV). The fraction emitted as β-particles is 3.6% (average energies are 0.445 MeV and 0.492 MeV), and the fraction emitted as γ-radiation is 1.1% (energy range, 0.01–1.27 MeV).

### High-throughput proliferation assay

Cells were plated using a Multidrop Combi liquid handler (Thermo Fisher Scientific) in at least 6 replicates at a single previously determined optimal cell density (range, 30–1600 cells/well) in a white 96-well plate with opaque walls (Corning). Plates were irradiated with a single dose of 137Cs or a continuous dose of 223RaCl2 with treatments delivered 24 hours after plating. Where indicated, hydroxyapatite (Sigma, H0252) was diluted (1:20,000) and added to the 100 μL of cell culture media at the time of seeding. At 9 to 11 days after irradiation, media were removed, and 50 μL of CellTiter-Glo reagent (50% solution in PBS; Promega) was added to each well (25). Relative luminescence units (RLU) were measured using an Envision multilabel plate reader (PerkinElmer) with a measurement time of 0.1 seconds. Luminescence signal is proportional to the amount of ATP present. The luminescence signal was plotted as a function of cell density, and a cell density within the linear range for luminescence (or growth) was selected to generate integral survival values for each cell line.

### Clonogenic survival

Cells were plated at appropriate dilutions, irradiated, and incubated for 7 to 21 days for colony formation. Colonies were fixed in a solution of acetic acid and methanol 1:3 (v/v) and stained with 0.5% (w/v) crystal violet as previously described (26). A colony was defined to consist of 50 cells or greater. Colonies were counted digitally using ImageJ software as described (27).

### Integral survival and relative biological effectiveness

The integral area under the curve (AUC) was estimated by trapezoidal approximation. The survival values for each trapezoid were multiplied by the dose interval, [f(X1) + f(X2)/2] × ΔX, and summed. To avoid differences in relative biological effectiveness (RBE) along the shape of the dose response curves (e.g., D50, D10, or Do), values were calculated as the ratio of (AUCγ/AUCα) for each cell line. To relate our estimates with RBE values reported by others, we also calculated RBE using D37.

### Microscopy

Tumor cells were plated in a 96-well half area high-content imaging glass bottom microplate (Corning) at a density of 500 to 2,000 cells/well. Wells were pretreated using poly-D-lysine (Sigma) 2 hours before cellular plating. Cells were then fixed and permeabilized using the Image-iT Fixation/Permeabilization Kit (Thermo Fisher Scientific). After fixation, the cytosol and nuclei were stained using actin green probes (green; Thermo Fisher Scientific) and DAPI (blue) or propidium iodide (red; Thermo Fisher Scientific), respectively. For each tumor cell line, at least 2 wells and 4 images per well (8 images in total) were captured at ×4 magnification using a Cytation 1 cell imaging multimode reader (BioTek). All images were processed manually using the ImageJ software. To measure the diameter of single cells (rather than clusters of cells), a threshold was set for the minimum and maximum pixel area size to exclude clusters of cells. Estimates of cytoplasmic and nuclear diameter were made by circular or ellipsoid fitting, outlining the chosen pixel area based on fluorescence intensity. The average cytoplasmic and nuclear diameters (the average of the major and minor axes in the case of ellipsoid fitting) of at least 100 cells per tumor cell line were calculated and their radii were used as input for microdosimetric calculations.

In the decay chain of 223Ra, a total of four high-energy daughter α-particles (219Rn, 215Po, 211Bi, and 211Po) and two beta decays (211Pb and 207Tl) are generated, with 207Pb as the final stable end product. We used the microdosimetry schema proposed by Roeske and Stinchcomb for calculations of absorbed α-particle dose, D, in Gy, for given α-particle energies, our source location, and individual target sizes (28). The values for the input quantities are calculated such that D = n\ x\ {z_1}$⁠, where n is the average number of hits to the target and {z_1}$ is the average of the single-hit specific energy (energy deposited per unit mass). {z_1}$values were either explicitly tabulated or obtained by linear interpolation for each cellular and nuclear radii pair. For the source-target geometry where the source is the medium outside the cell, its volume was taken to be one cm3 with the target cell at the center. This choice for the source volume allows for using the cumulated activity per cm3 such that S is equal to the absorbed dose per unit of cumulated activity (Gy•cm3/Bq•s). We used S = {10^{ - 12}}\pi ( {{z_1}} )[ {{R_\alpha }{r^2} + ( {\frac{2}{3}} )\{ {{r^3} + {{( {{s^2} - {r^2}} )}^{3/2}} - {s^3}} \}} ]$⁠, where {R_\alpha }$is the range of the α-particle, and s and r are cellular and nuclear radii, respectively. In our system, all of our cells adhered to the bottom of the well so we applied a correction of 0.5 to the S-value to take into account the exclusively hemispheric irradiation (only from the top). In the presence of hydroxyapatite, the correction was not required due to the spherical distribution of irradiation; the cells are fully immersed in the matrix. For cells that had an ellipsoid shape, a 0.9 correction was applied. This correction is derived from the relationship between S and electron energy for an ellipsoid geometry and is based on an average particle range of ∼60 μm and its corresponding electron energy. We used D = A\ x\ S$ to calculate the absorbed dose for each α-particle. The cumulated activity for 223Ra at initial activity {A_0}$was calculated using A = {A_0}{e^{ - \lambda t}}$⁠, where \lambda = \ - 0.693/{T_{{\rm{half \hbox{-} life}}}}$⁠, and integrated for the duration of exposure. The cumulated activity for each α-particle emitting daughter was calculated using {A_2} = ( {\frac{{{\lambda _2}}}{{{\lambda _2} - \ {\lambda _1}}}} ){A_0}( {{e^{ - {\lambda _1}t}} - \ {e^{ - {\lambda _1}t}}} )$⁠, where {\lambda _2}$is the decay constant for the daughter nuclide and {\lambda _1}$ is the decay constant of 223Ra. We set the decay constant {\lambda _1}$to 223Ra for all daughters because the activity of the daughters is short-lived and dependent on the half-life of 223Ra, which is significantly greater than 219Rn, 215Po, and 211Bi (3.97 s, 1.78 ms, and 2.14 min, respectively). In addition, although there is some initial activity attributed to the daughters, we assume that the daughter nuclides have zero activity at zero time. Because the half-life of 223Ra >> 219Rn, 215Po, and 211Bi, the activity contributed by the daughters immediately upon cellular irradiation is negligible compared with the overall accumulated activity. ### Variant generation in lentiviral vectors We performed mutagenesis in three steps: PCR, in vitro recombination and transformation. Briefly, the gene ORF was PCR amplified by using primers that contain incorporated mutated sequence. Fragments were then transferred directly to the destination vector (pLEX306 or pLEX307) by LR recombination (Invitrogen) and the constructs were transformed into competent cells. The discontinuity at the mutation site was repaired by endogenous bacterial repair mechanism. After virus infection (multiplicity > 1), BEAS-2B cells were selected in the presence of 1 μg/mL puromycin. ### Western blot analysis Whole-cell lysates were made using M-PER lysis buffer (Thermo Fisher Scientific). Proteins were separated on 4% to 12% bis-tris SDS-PAGE gels with MOPs buffer and transferred onto 0.45 μmol/L nitrocellulose (Thermo Fisher Scientific). Blots were developed with ECL Prime Western blotting detection reagent (Amersham/GE Healthcare). Anti-PIK3CA (clone C73F8, #4249, 1:2,000), anti-AKT (#9272, 1:2,000), antiphospho-S473-AKT (#9271, 1:1,000), anti-HER2 (clone D8F12, #4290, 1:2,000), antiphospho-Y1248-HER2 (#2247, 1:1,500), and anti-β-actin (clone 13E5, #4970, 1:5,000) were from Cell Signaling Technology. ### Information-based association score The association between genomic alterations [e.g., mutations or somatic copy-number alteration (SCNA)] or single-sample gene set enrichment analysis (ssGSEA) profiles for each gene set and the radiation response profile was determined using the information coefficient (IC; refs. 25, 29, 30). ### Genetic data Cancer cell lines were profiled at the genomic level and processed as described in detail (24). The processed data are available for download at http://www.broadinstitute.org/ccle. Briefly, mutation information was obtained by using massively parallel sequencing of exomes. Genotypes were transformed to categorical values (mutation = 1, no mutation = 0) and were used as input to compute the IC. Genotyping/copy-number analysis was performed using Affymetrix Genome-Wide Human SNP Array 6.0. Raw Affymetrix CEL files were converted to a single value for each probe set representing an SNP allele or a copy-number probe using a GenePattern pipeline (31) and hg18 Affymetrix probe annotations. Copy numbers were then inferred based upon estimating probe set specific linear calibration curves, followed by normalization by the most similar HapMap normal samples. Segmentation of normalized log2 ratios [specifically, log2(CN/2)] was performed using the circular binary segmentation algorithm (32), followed by median centering of the segment values to a value of zero in each sample. Next, quality checking of each array was performed, including visual inspection of the array pseudo-images, probe-to-probe noise variation between copy-number values, confidence levels of Birdseed genotyping calls, and appropriate segmentation of the copy-number profiles (33). Finally, the Genomic Identification of Significant Targets in Cancer (GISTIC) algorithm was used to identify focal regions of copy-number alterations in individual samples (34). A gene-level copy number was also generated, defined as the maximum absolute segmented value between the gene's coordinates, and calculated for all genes using the hg18 coordinates provided by the refFlat and wgRna databases from the UCSC Genome Browser (http://hgdownload.cse.ucsc.edu/goldenPath/hg18/database/). Separate binary variables representing amplifications (above 0.7) and deletions (below −0.7) were generated based on the GISTIC gene-level copy-number output described above. These binary amplification/deletion variables for each gene were used as input to compute the IC against the radiation sensitivity phenotype. mRNA gene expression was measured by RNA-seq. RPKM values were used as input to calculate the ssGSEA enrichment scores based on the weighted difference of the empirical cumulative distribution functions of the genes in the set relative to the genes not included in an individual set (35). The result is a single score per cell line per gene set, transforming the original data set into a more interpretable higher-level description. Gene sets were obtained from the C2 subcollection of the Molecular Signatures Database (MSigDB), an additional collection of oncogenic signatures, and other cancer-related gene sets curated from the literature, resulting in a data set that has 5,826 pathway profiles for each sample (36). ssGSEA values were used as input to compute the IC. The nominal P values for the information-based association metric scores between the genetic parameters (alterations or ssGSEA scores) and radiation response scores were estimated using an empirical permutation test. ### Development and validation of a high-content α-particle irradiation survival platform We profiled 28 cancer cell lines comprising 10 cancer types by modifying an established high-throughput profiling platform for cellular survival after γ irradiation (7, 25). To profile the response of cancer cells to α-particles in a format amenable to high-throughput profiling, we first determined the proliferating fraction (mean RLU at dose x/mean RLU of control) as a function of dose for all cell lines using optimized growth measurements in a 96-well plate (Fig. 1A). We calculated the integral survival value for each cell line and compared it with the corresponding value after single fraction exposure to sparsely ionizing γ (137Cs) irradiation (Supplementary Fig. S1). We showed that although cell lines that were more resistant to γ irradiation were generally more likely to be less sensitive to α-particles, 223Ra was more effective in inducing cellular growth delay and/or death in all 28 cell lines (Fig. 1B and C). There were notable exceptions to the overall correlation above and below the trendline. These, in part, included less sensitive cells, including two classes of small cell lung carcinoma with neuroepithelial and mesenchymal differentiation (DMS53 and DMS114) and a breast carcinoma cell line with ERBB2 amplification, AR overexpression, and a canonical PIK3CA mutation (HCC202). Interestingly, these and other relatively more resistant cells had a longer cellular doubling time (Pearson r = 0.60, R2 = 0.35, and P = 0.03; Supplementary Fig. S2). We tested the impact of the duration of exposure to α-particle on the relative sensitivity of cells with short (HSC4, RERFLCAI, and HCT15) or longer doubling times (LNCaP and BT474). Reduction of the exposure time to 24 hours, with time to readout at 8 to 10 days after washout, did not significantly alter the relative sensitivity of the five cell lines (Supplementary Fig. S3). These data indicate that the relationship between doubling time and radiation sensitivity is not merely related to the experimental design and that cellular repopulation is not a major determinant of decreased sensitivity up to 10 days after radiation treatments. Figure 1. High-throughput profiling of cellular survival after α-particle irradiation. A, The surviving fraction after treatment with 223Ra for all 28 cell lines. Dose represents the initial activity of 223Ra. Data are expressed as the means ± SEM. B, Integral survival of dose (in kBq) was calculated for each cell line profiled by the high-throughput profiling platform after treatment with 223Ra and was compared with integral survival values of the corresponding cell lines after treatment with γ-irradiation (137Cs). Data points represent the mean of at least two experiments. The diagonal represents the iso-sensitivity delimiter. C, Integral survival of dose (in kBq) after 223Ra and 137Cs treatments is plotted as box and rotated kernel density plots. *, P < 0.001, Wilcoxon test for paired values. D, Integral survival of dose (in Gy) after 223Ra treatment was measured by the high-throughput profiling (HTP) platform and by colony formation assay (CFA), was analyzed by linear regression, and the R2 value was calculated. Data points represent the mean of at least two experiments. Figure 1. High-throughput profiling of cellular survival after α-particle irradiation. A, The surviving fraction after treatment with 223Ra for all 28 cell lines. Dose represents the initial activity of 223Ra. Data are expressed as the means ± SEM. B, Integral survival of dose (in kBq) was calculated for each cell line profiled by the high-throughput profiling platform after treatment with 223Ra and was compared with integral survival values of the corresponding cell lines after treatment with γ-irradiation (137Cs). Data points represent the mean of at least two experiments. The diagonal represents the iso-sensitivity delimiter. C, Integral survival of dose (in kBq) after 223Ra and 137Cs treatments is plotted as box and rotated kernel density plots. *, P < 0.001, Wilcoxon test for paired values. D, Integral survival of dose (in Gy) after 223Ra treatment was measured by the high-throughput profiling (HTP) platform and by colony formation assay (CFA), was analyzed by linear regression, and the R2 value was calculated. Data points represent the mean of at least two experiments. Close modal We next examined whether the high-throughput platform correlated with clonogenic survival following exposure to 223Ra. Mean integral survival values for 12 cell lines (for each cell line, n ≥ 2) was calculated and compared with values from the clonogenic assay (Fig. 1D). Proliferation and colony integral survival values were significantly correlated, with Pearson r = 0.78, R2 = 0.61, and P = 0.003. Therefore, the high-throughput platform accurately profiles cancer cell lines for response to radium and demonstrated significant variation in survival across and within several cancer types. ### α-Particle microdosimetry and cellular morphology In our experimental system, our targets (nuclei and cytoplasms) receive dose from medium outside of the cell from 223Ra and its daughters over 9 to 11 days of continuous exposure. In this setting of cells irradiated in a uniform solution of α-particles, the activity of the medium alone is insufficient to estimate the absorbed dose for each tumor cell line. First, the average absorbed dose across a panel of cells with distinct cellular and nuclear dimensions, and therefore target size, is likely to vary. Second, because the range of the α-particle is on the same order of magnitude as the diameter of the target (e.g., the nucleus), not all cells of a particular cell line will receive the same dose. Therefore, a dosimetric framework that can account for intercellular and intracellular variation in absorbed dose estimates is essential to interpret biological results. To accurately estimate the average absorbed dose, we first calculated the average specific energy deposited to each tumor cell based on the individual nuclear and cytoplasmic dimensions (i.e., radii). We used fluorescence capture of nuclear and cytoplasmic staining to calculate the dimensions of each tumor cell line (Fig. 2A). Although we found some variation in the radial dimension and ploidy of tumor cell nuclei, the variation was significantly greater for cytoplasms compared with nuclei, σ2 = 3.2 vs. 0.7, respectively (F test, P ≤ 0.001; Fig. 2B). With these parameters as input, we calculated the average absorbed dose in each cell line from all α-particle emissions (223Ra, 219Rn, 215Po, and 211Bi; Fig. 2C). We showed that the activity (in kBq) is highly correlated with the absorbed dose (in Gy) across the 28 cell lines (R2 = 0.97, P ≤ 0.001; Fig. 2D). Linear regression showed a slope of 2.4 ± 0.1, reflecting the lower areas under the curve after adjusting for absorbed dose. Figure 2. High-content cellular microdosimetery for 223Ra treatments. A, The cytosol and nuclei for each cell line were stained green and blue, respectively. Magnification, ×20. Scale bars, 100 μm. B, Histogram and probability density function of measured radii for all 28 cell lines. C, Schematic depiction of cells exposed to 223Ra and α-particle products along its decay chain. The range for the path lengths of the four particles produced is shown. t = 10 days, the average number of days of α-particle incubation with cells. D, Integral survival calculated from 223Ra initial radioactivity measurements (kBq) was plotted against values calculated from 223Ra absorbed dose (Gy). Figure 2. High-content cellular microdosimetery for 223Ra treatments. A, The cytosol and nuclei for each cell line were stained green and blue, respectively. Magnification, ×20. Scale bars, 100 μm. B, Histogram and probability density function of measured radii for all 28 cell lines. C, Schematic depiction of cells exposed to 223Ra and α-particle products along its decay chain. The range for the path lengths of the four particles produced is shown. t = 10 days, the average number of days of α-particle incubation with cells. D, Integral survival calculated from 223Ra initial radioactivity measurements (kBq) was plotted against values calculated from 223Ra absorbed dose (Gy). Close modal We also correlated integral survival values using the calculated absorbed dose with cellular parameters across the 28 cell lines. Cells with larger cytoplasmic, but not nuclear, radii appeared to correlate with integral survival (Supplementary Fig. S4a and S4b). Specifically, cells with larger cytoplasmic radii were more likely to be sensitive to α-particles. Correspondingly, nuclear-to-cytoplasmic ratios were also associated with radiation sensitivity (R2 = 0.5, P < 0.0001; Supplementary Fig. S4c). These results suggested that cellular shape and morphology may affect cancer survival after α-particle irradiation. ### The osteomimetic hydroxyapatite and the cellular response to α-particles Hydroxyapatite is a calcium phosphate similar to the human bone in morphology and composition insofar as it has a hexagonal structure and a similar stoichiometric calcium-to-phosphate ratio (37). Hydroxyapatite has previously been shown to induce osteogenic differentiation from cellular precursors (38) and alter the transcriptome and proteome of cocultured cells (39). Because 223Ra exerts its effect by incorporation into the bone, we sought to assess the impact of hydroxyapatite on cellular sensitivity to α-particles using a coincubation method (Fig. 3A). First, we measured the cellular survival as a function of the initial activity of 223Ra in the medium without or with hydroxyapatite (Fig. 3B). Both the GI50 (compare Fig. 1A and Fig. 3B) and the integral survival values were significantly reduced by the addition of hydroxyapatite (Fig. 3C). This suggested that 223Ra effectively bound to the hydroxyapatite matrix and was more effective in this source geometry per unit of activity than as an element in suspension. We then calculated the absorbed dose and considered the source geometry after exchange between calcium and 223Ra. In this system, irradiation occurs from both the top and bottom of the cell (no hemispheric correction; see Materials and Methods). After adjusting for the source geometry, there was no difference in the overall sensitivity of cells to α-particles without or with hydroxyapatite (Fig. 3D). Figure 3. Hydroxyapatite and cellular survival after α-particle irradiation. A, Schematic depiction of cells cocultured with hydroxyapatite. B, The surviving fraction of cells cocultured with hydroxyapatite after 223Ra treatment. Dose represents the initial activity of 223Ra. Data are expressed as the means ± SEM. C, Integral survival values calculated from 223Ra initial radioactivity measurements (kBq) were plotted with and without hydroxyapatite for each cell line. D, Integral survival values calculated from 223Ra absorbed dose (Gy) measurements were plotted with and without hydroxyapatite for each cell line. Figure 3. Hydroxyapatite and cellular survival after α-particle irradiation. A, Schematic depiction of cells cocultured with hydroxyapatite. B, The surviving fraction of cells cocultured with hydroxyapatite after 223Ra treatment. Dose represents the initial activity of 223Ra. Data are expressed as the means ± SEM. C, Integral survival values calculated from 223Ra initial radioactivity measurements (kBq) were plotted with and without hydroxyapatite for each cell line. D, Integral survival values calculated from 223Ra absorbed dose (Gy) measurements were plotted with and without hydroxyapatite for each cell line. Close modal ### Relative biological effectiveness of α-particles We also compared the efficacy of α-particle irradiation to that of sparsely ionizing radiation (137Cs). For each cell line, we used {\rm{RBE}} = \frac{{{\rm{AU}}{{\rm{C}}_\alpha }}}{{{\rm{AU}}{{\rm{C}}_\gamma }}}$ to calculate the relative biological effectiveness of α-particles relative to γ irradiation. The median RBE, calculated using AUC estimates, was 10.4 (IQR: 8.4–14.3). The median RBE, calculated using D37, was 9.7 (IQR: 4.5–12). These results indicate that α-particles are significantly more effective in tumor cell killing than γ-rays.

Unlike α-particle emissions, sparsely ionizing radiation causes DNA damage mostly through intermediary reactive oxygen species. NRF2 and its binding partner KEAP1 are key regulators of oxidative stress response (40, 41). We previously identified mutations in the NRF2 pathway as some of the most highly correlated with resistance to γ irradiation (7, 25). Accordingly, we sought to assess the relative importance of the oxidative response pathway in regulating survival after α-particle irradiation. Cell lines with NRF2 (NFE2L2) or KEAP1 mutations were significantly more sensitive to α-particle treatments than γ irradiation and were among some of the most sensitive cells to these treatments (Fig. 4A; Supplementary Table S1).

Figure 4.

Nrf2-mutant cells are effectively treated with α-particle irradiation. A, Integral survival values of cell lines with Nrf2 (NFE2L2) or KEAP1 mutations after 223Ra and 137Cs treatments are shown. Black bars, median values. B, BEAS-2B cells stably infected with vector alone (φ) or vector expressing Nrf2 alleles were profiled for Nrf2 protein level by immunoblot. C, BEAS-2B cells from B were irradiated with either 223Ra or 137Cs. Data points represent mean ± SE and are representative of at least three experiments. P < 0.05 and q value (false discovery rate calculated using the two-stage step-up method) <0.01 were considered statistically significant and are denoted by an asterisk.

Figure 4.

Nrf2-mutant cells are effectively treated with α-particle irradiation. A, Integral survival values of cell lines with Nrf2 (NFE2L2) or KEAP1 mutations after 223Ra and 137Cs treatments are shown. Black bars, median values. B, BEAS-2B cells stably infected with vector alone (φ) or vector expressing Nrf2 alleles were profiled for Nrf2 protein level by immunoblot. C, BEAS-2B cells from B were irradiated with either 223Ra or 137Cs. Data points represent mean ± SE and are representative of at least three experiments. P < 0.05 and q value (false discovery rate calculated using the two-stage step-up method) <0.01 were considered statistically significant and are denoted by an asterisk.

Close modal

To study the effects of NRF2 directly, we expressed NRF2 wild-type and T80K in an immortalized human bronchial epithelial cell line (BEAS-2B). Mutation T80K has previously been shown to abrogate binding to KEAP1, resulting in activation of the NRF2 pathway (42). We showed that overexpression T80K led to an increase in NRF2 protein level, followed by wild-type and then vector control cells (Fig. 4B). Correspondingly, NRF2 T80K proteins, and to a lesser extent wild-type, significantly increased the resistance of BEAS-2B cells to γ irradiation from a 137Cs source (Fig. 4C). In contrast, BEAS-2B cells expressing either T80K or wild-type NRF2 did not demonstrate significantly different survival measurements compared with vector alone cells after α-particle irradiation. We note that T80K mutations, and to a lesser extent wild-type NRF2, contributed to marginally improved survival after α-particle irradiation compared with vector control despite not achieving statistical significance (Supplementary Fig. S5).

Altogether, these results quantify the relative potency of α-particles and demonstrate that they can effectively lethally damage cells that are the most resistant to sparsely ionizing radiation at much lower absorbed radiation values.

### Integrative genetic profiling of the cellular response to α-particles

We observed significant variation in survival across cell lines, on the order of a 32-fold difference between the most sensitive and resistant cells (Fig. 5A). The variation in α-particle response juxtaposed with the genetic heterogeneity of the profiled cell lines suggested that genetic parameters may regulate response to α-particle irradiation. We sought to identify these putative associations. We used ssGSEA projections as a gene set identification tool to find genetic pathways that are differentially correlated with radiation response (see Materials and Methods; Fig. 5B; refs. 25, 35). We also identified gene mutations and SCNA that correlated with radiation sensitivity across all cell lines (Fig. 5C and D). The recent availability of whole-exome sequencing data from the CCLE significantly broadened our association analysis from ∼1,600 genes to ∼19,000 (43).

Figure 5.

Gene-expression changes, SCNA, and mutations associated with cellular survival after α-particle irradiation. A, Histogram and probability density function of calculated integral survival values of 28 cell lines. B, ssGSEA identifies gene sets that correlate with radiation sensitivity and resistance. Heat map of ssGSEA scores (red, positive; blue, negative). Top gene sets, organized by biological processes, are shown. C, A subset of the top SCNAs that correlate with radiation resistance and sensitivity is organized by chromosomal positions. Red and blue bars represent an SCNA in the corresponding gene that is associated with resistance and sensitivity, respectively. D, A subset of the top genes that, when mutated, were associated with resistance or sensitivity. Red and blue bars represent a mutation in the corresponding gene that is associated with resistance and sensitivity, respectively. Heat map of integral survival (red, resistant; blue, sensitive) is parallel to the representations in B, C, and D.

Figure 5.

Gene-expression changes, SCNA, and mutations associated with cellular survival after α-particle irradiation. A, Histogram and probability density function of calculated integral survival values of 28 cell lines. B, ssGSEA identifies gene sets that correlate with radiation sensitivity and resistance. Heat map of ssGSEA scores (red, positive; blue, negative). Top gene sets, organized by biological processes, are shown. C, A subset of the top SCNAs that correlate with radiation resistance and sensitivity is organized by chromosomal positions. Red and blue bars represent an SCNA in the corresponding gene that is associated with resistance and sensitivity, respectively. D, A subset of the top genes that, when mutated, were associated with resistance or sensitivity. Red and blue bars represent a mutation in the corresponding gene that is associated with resistance and sensitivity, respectively. Heat map of integral survival (red, resistant; blue, sensitive) is parallel to the representations in B, C, and D.

Close modal

We compared the profiles of each gene set with the α-particle radiation response scores (integral survival). The ssGSEA scores are displayed in a heat map with the top gene sets that correlate with radiation survival organized by biological annotation (Fig. 5B; Supplementary Table S2). The top gene sets that correlated with reduced radiation sensitivity revealed pathways associated with breast cancer, cellular signaling and hypoxia. Within the breast cancer category, estrogen signaling, the luminal B subtype and ERBB2 signaling were associated with reduced sensitivity. Interestingly, several individual gene sets within the breast category demonstrate that subtypes with a propensity to metastasize to the bone are less likely to be sensitive to α-particles (e.g., SMID_RELAPSE_IN_BONE_UP).

To assess the association of individual SCNA with radiation response, we correlated alterations with radiation survival using the IC (Fig. 5C; Supplementary Table S3). Consistent with the association in ERBB2 cell signaling by ssGSEA, ERBB2 amplification was also associated with decreased sensitivity to α-particle treatments. In fact, ERBB2 amplification scored second out of 46,637 potential gene-level SCNA associations (IC = 0.592, P ≤ 0.001). Consistent with these results, overexpression of ERBB2 has previously been shown to confer therapeutic resistance, and trastuzumab, a monoclonal antibody that interferes with ErbB2, sensitizes ErbB2-expressing cells to sparsely ionizing radiation (44).

### PIK3CA variants reduced cellular sensitivity to α-particle irradiation

To assess the association of individual mutations with radiation response, we correlated mutations with radiation survival using the IC (Fig. 5D; Supplementary Table S4). We associated whole-exome data, which included sequencing data from 18,750 genes, with cellular survival to 223Ra. We identified mutations in PIK3CA as one of the top genes associated with decreased sensitivity to irradiation (IC = 0.402, P = 0.03), which ranked 29th overall. We mapped the individual PIK3CA mutations on a linear protein coordinate and delineated its domains (Fig. 6A). E545K, E542K, and H1047R were well represented in the profiled cell lines, implicating functionally relevant and frequent mutations in PIK3CA (45).

Figure 6.

PIK3CA activation mutations confer decreased sensitivity to α-particles. A, A map of the PIK3CA mutations identified in the profiled cell lines displayed on the protein and its domains. B, BEAS-2B cells stably infected with vector alone (φ) or vector expressing PIK3CA alleles were profiled for PI3K/AKT pathway activity by immunoblot. C, The cytosol and nuclei for each cell line were stained green and red, respectively. D, Vector alone (φ) or vector expressing PIK3CA alleles were profiled for survival from 223Ra irradiation. Data are expressed as the AUC and represent the mean ± SEM of at least three independent experiments. Vector alone (φ), vector expressing PIK3CA wild-type, or H1047R cells were profiled using either the high-content proliferation method (E) or by colony formation assay (F and G). Data points represent mean ± SE and are representative of at least three experiments. P < 0.05 and q value (false discovery rate calculated using the two-stage step-up method) <0.01 were considered statistically significant and are denoted by an asterisk. Representative irradiated colonies in G are from a dose of 0.076 Gy of 223Ra.

Figure 6.

PIK3CA activation mutations confer decreased sensitivity to α-particles. A, A map of the PIK3CA mutations identified in the profiled cell lines displayed on the protein and its domains. B, BEAS-2B cells stably infected with vector alone (φ) or vector expressing PIK3CA alleles were profiled for PI3K/AKT pathway activity by immunoblot. C, The cytosol and nuclei for each cell line were stained green and red, respectively. D, Vector alone (φ) or vector expressing PIK3CA alleles were profiled for survival from 223Ra irradiation. Data are expressed as the AUC and represent the mean ± SEM of at least three independent experiments. Vector alone (φ), vector expressing PIK3CA wild-type, or H1047R cells were profiled using either the high-content proliferation method (E) or by colony formation assay (F and G). Data points represent mean ± SE and are representative of at least three experiments. P < 0.05 and q value (false discovery rate calculated using the two-stage step-up method) <0.01 were considered statistically significant and are denoted by an asterisk. Representative irradiated colonies in G are from a dose of 0.076 Gy of 223Ra.

Close modal

We sought to examine the impact of activating mutations in PIK3CA on cellular survival after α-particle irradiation without the confounding effects of varied background genetic alterations in the profiled human cancer derived cell lines. To achieve this, we expressed PIK3CA E545K and H1047R in the same genetically defined, immortalized human bronchial epithelial cell line, BEAS-2B. PIK3CA variants and not wild-type led to constitutive activation of the PI3K–AKT signaling pathway (Fig. 6b). PIK3CA variant expressing cells had a distinct cellular morphology compared with wild-type expressing and vector control cells (Fig. 6C). Importantly, although the nuclear-to-cytoplasmic ratios did not vary substantially based on the genotype, we accounted for the larger nuclear and cytoplasmic radii in PIK3CA mutant and, to a lesser extent, wild-type cells compared with cells with vector alone in our calculation of the absorbed dose of 223Ra. Both PIK3CA E545K and H1047R, but not wild-type, enhanced cellular survival to α-particle irradiation (Fig. 6D). We evaluated H1047R using both proliferation profiling (Fig. 6E) and colony formation assays to confirm these findings (Fig. 6F and G). These results indicated that activating mutations in PIK3CA decrease the sensitivity of cells to α-particle irradiation.

We developed and benchmarked a high-content platform that measures 223Ra radiation survival across a diverse panel of tumor cells. The platform integrates fluorescence microscopy, individual cellular microdosimetry, and computational approaches that span the preclinical experimental continuum from measuring radiation sensitivity to associating cancer genetic data with treatment responses. We used this platform to profile 28 cancer cell lines for response to 223Ra and estimated the absorbed dose to each cell line. Critically, the dose range in which we observed a dose response to α-particle treatments across multiple cell lines (the steep aspect of the survival curves) significantly overlapped with clinical estimates of the mean absorbed dose to metastatic tumors in bone after Ra223 treatments (0.2–1.9 Gy; refs. 46–48). These findings, coupled with the expected significant stochastic variations of energy deposited within small targets (29), suggest that survival measurements from our platform are likely to be clinically relevant.

We showed that α-particle treatments were significantly more effective than sparsely ionizing radiation with an RBE of ∼10, which is 2-fold higher than its previously estimated magnitude (13, 14). These results are consistent with the proposed direct action of the incident α-particle on DNA and the attenuated cross-resistance to γ irradiation with α-particles compared with the converse (49). Coupled with approaches to estimate the RBE for normal tissues (50), more individualized estimates of RBE based on genomic biomarkers could guide clinical dose schedules for α-emitting radionuclides.

Despite the increased sensitivity of cells to α-particles, the distribution of integral survival measurement across the panel of cells suggested significant underlying cellular and genetic diversity. Specifically, a trend in association was observed between nuclear-to-cytoplasmic ratios and responses to α-particles. This associations appear to be mainly driven by cytoplasmic size, with larger cells demonstrating higher sensitivity to treatments. There is evidence to support a modest contribution toward DNA damage from cytoplasmic irradiation (51, 52). However, the magnitude of the effect that we observed suggests a more substantial association between cytoplasmic irradiation and survival. It is unclear whether this association is regulated by energy deposition by an α-particle produced outside of the cell or whether continuous exposure with 223Ra allows for intracellular sequestration via exchanges with similar elements (i.e., calcium, magnesium, iron, and/or copper). For example, there is some evidence for sequestration of 223Ra by intracellular ferritin (53).

We studied the effects of bony matrix on cellular sensitivity. We found that the radioactivity required to effect iso-sensitivity was significantly less when cells were grown in the presence of bone-like hydroxyapatite matrix in vitro, indicating that there was exchange between the hydroxyapatite (calcium) and 223Ra. However, adjustment for the source geometry resulted in a corrected absorbed dose of irradiation that indicated no significant differences in sensitivity. These results indicate that our measurement of cellular sensitivity is not altered when cells are placed in a matrix that mimics bone. The implication of these findings is that candidate biomarkers identified through our profiling effort can potentially direct the use of α-particles to treat tumors that are located in the bone or the viscera.

We identified several gene-expression set determinants of response to α-particles. The identification of breast cancer cells overall and mainly those with a propensity to travel to bone, the luminal B subtype (54), as markers of decreased sensitivity may inform clinical studies of 223Ra in patients with metastatic breast cancer. We also identified genetic alterations that can potentially have predictive capacity by identifying the likelihood of response to treatments. A subset of these alterations (e.g., ERBB2 amplification) can potentially guide combinatorial therapeutic strategies because these alterations both conferred decreased sensitivity and are targets of approved drugs. Lastly, the oxygen enhancement ratio (OER) of high linear energy transfer (LET) particles is predicted to be substantially less than sparsely ionizing radiation. Nonetheless, the LET of α-particles is in the range of 60 to 110 keV/μm, resulting in an OER that remains >1 (55). This, coupled with the biological effects of hypoxia beyond oxygen fixation of DNA damage, suggests that hypoxic tumors may remain relatively more resistant to α-particles than their nonhypoxic counterparts as indicated in our gene set associations.

Importantly, we demonstrated that cells that are among the most resistant to γ radiation can be effectively treated with α-particles. The largest radiation cell line profiling effort conducted to date revealed that cells with alterations in oxidative stress response, namely, NRF2 and its binding partner KEAP1, are highly associated with γ radiation resistance (7). Our results, in both cancer cell lines and immortalized human bronchial epithelial cells made to express an activating allele of NRF2, indicate that α-particles can effectively kill these cells. These results can affect future strategies of radiation dose escalation in tumors with a preponderance of NRF2/KEAP1 alterations. Those include tumors of the head and neck, esophagus, lung, and bladder (56).

We validated our genetic biomarker platform by demonstrating that PIK3CA-activating mutations confer decreased sensitivity to α-particles. This finding is critical because it demonstrates that some cancer cells, despite continuous and uniform irradiation, show decreased sensitivity to high LET irradiation. Moreover, these results demonstrate that survival is guided by genetic alterations, indicating that biologically guided radiotherapy will also be relevant for α-particle treatments. The high frequency of PI3K pathway alterations across several cancer types presents an opportunity for the combinatorial targeting of tumors with both α-particle and PI3K inhibitors.

In summary, we have established a platform designed for biomarker and target identification to α-particle treatments. Our strategies could guide appropriate patient selection for these treatments and result in the improved clinical outcomes for patients with tumors that are putatively the most resistant to sparsely ionizing radiation treatments.

M.E. Abazeed reports receiving a commercial research grant from Bayer AG, other commercial research support from Siemens Healthcare Solutions USA, and honoraria from the speakers bureau of Bayer AG. No potential conflicts of interest were disclosed by the other authors.

Conception and design: B.D. Yard, G. Siemeister, U.B. Hagemann, M.E. Abazeed

Development of methodology: B.D. Yard, K. Bannik, M.E. Abazeed

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.D. Yard, P. Gopal, M.E. Abazeed

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.D. Yard, P. Gopal, G. Siemeister, M.E. Abazeed

Writing, review, and/or revision of the manuscript: P. Gopal, G. Siemeister, U.B. Hagemann, M.E. Abazeed

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): U.B. Hagemann, M.E. Abazeed

Study supervision: M.E. Abazeed

M.E. Abazeed was supported by NIH KL2TR0002547, NIH R37CA222294, and VeloSano.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Bernier
J
,
Hall
EJ
,
Giaccia
A
.
Radiation oncology: a century of achievements
.
Nat Rev Cancer
2004
;
4
:
737
47
.
2.
Brahme
A.
Accurate description of the cell survival and biological effect at low and high doses and LET's
.
2011
;
52
:
389
407
.
3.
Howard-Flanders
P.
Physical and chemical mechanisms in the injury of cells of ionizing radiations
.
1958
;
6
:
553
603
.
4.
DT.
Initial events in the cellular effects of ionizing radiations: clustered damage in DNA
.
1994
;
65
:
7
17
.
5.
DT.
The initial physical damage produced by ionizing radiations
.
1989
;
56
:
623
34
.
6.
Hill
MA
.
Radiation damage to DNA: the importance of track structure
.
1999
;
31
:
15
23
.
7.
Yard
BD
,
DJ
,
Chie
EK
,
Tamayo
P
,
Battaglia
JS
,
Gopal
P
, et al
A genetic basis for the variation in the vulnerability of cancer to DNA damage
.
Nat Commun
2016
;
7
:
11428
.
8.
Hanna
TP
,
Shafiq
J
,
Delaney
GP
,
Vinod
SK
,
Thompson
SR
,
Barton
MB
.
The population benefit of evidence-based radiotherapy: 5-year local control and overall survival benefits
.
2018
;
126
:
191
7
.
9.
Kulakowski
A.
The contribution of Marie Sklodowska-Curie to the development of modern oncology
.
Anal Bioanal Chem
2011
;
400
:
1583
6
.
10.
DT
,
Munson
RJ
,
Thacker
J
,
Cox
R
.
Mutation and inactivation of cultured mammalian cells exposed to beams of accelerated heavy ions. IV. Biophysical interpretation
.
Int J Radiat Biol Relat Stud Phys Chem Med
1980
;
37
:
135
67
.
11.
Franken
NA
,
ten Cate
R
,
Krawczyk
PM
,
Stap
J
,
Haveman
J
,
Aten
J
, et al
Comparison of RBE values of high-LET alpha-particles for the induction of DNA-DSBs, chromosome aberrations and cell reproductive death
.
2011
;
6
:
64
.
12.
Franken
NA
,
Hovingh
S
,
Ten Cate
R
,
Krawczyk
P
,
Stap
J
,
Hoebe
R
, et al
Relative biological effectiveness of high linear energy transfer alpha-particles for the induction of DNA-double-strand breaks, chromosome aberrations and reproductive cell death in SW-1573 lung tumour cells
.
Oncol Rep
2012
;
27
:
769
74
.
13.
Sgouros
G
,
Roeske
JC
,
McDevitt
MR
,
Palm
S
,
Allen
BJ
,
Fisher
DR
, et al
MIRD Pamphlet No. 22 (abridged): radiobiology and dosimetry of alpha-particle emitters for targeted radionuclide therapy
.
J Nucl Med
2010
;
51
:
311
28
.
14.
Feinendegen LE
MJ.
Meeting report: Alpha-emitters for medical therapy—Workshop of the United States Department of Energy
.
1997
,
:
. p.
195
201
.
15.
Nilsson
S
,
Franzén
L
,
Parker
C
,
Tyrrell
C
,
Blom
R
,
Tennvall
J
, et al
Bone-targeted radium-223 in symptomatic, hormone-refractory prostate cancer: a randomised, multicentre, placebo-controlled phase II study
.
Lancet Oncol
2007
;
8
:
587
94
.
16.
Zustovich
F.
,
Barsanti
R.
Targeted alpha therapies for the treatment of bone metastases
.
Int J Mol Sci
2017
;
19
(1). doi: 10.3390/ijms19010074.
17.
Parker
C
,
Nilsson
S
,
Heinrich
D
,
Helle
SI
,
O'Sullivan
JM
,
Fosså
SD
, et al
Alpha emitter radium-223 and survival in metastatic prostate cancer
.
N Engl J Med
2013
;
369
:
213
23
.
18.
Poty
S
,
Francesconi
LC
,
McDevitt
MR
,
Morris
MJ
,
Lewis
JS
.
.
J Nucl Med
2018
;
59
:
878
84
.
19.
Hagemann
UB
,
Wickstroem
K
,
Wang
E
,
Shea
AO
,
Sponheim
K
,
Karlsson
J
, et al
In vitro and in vivo efficacy of a novel CD33-targeted thorium-227 conjugate for the treatment of acute myeloid leukemia
.
Mol Cancer Ther
2016
;
15
:
2422
31
.
20.
Hagemann
UB
,
Mihaylova
D
,
Uran
SR
,
Borrebaek
J
,
Grant
D
,
Bjerke
RM
, et al
Targeted alpha therapy using a novel CD70 targeted thorium-227 conjugate in in vitro and in vivo models of renal cell carcinoma
.
Oncotarget
2017
;
8
:
56311
26
.
21.
Hagemann
UB
,
Ellingsen
C
,
Schuhmacher
J
,
Kristian
A
,
Mobergslien
A
,
Cruciani
V
, et al
Mesothelin-targeted thorium-227 conjugate (MSLN-TTC): preclinical evaluation of a new targeted alpha therapy for mesothelin-positive cancers
.
Clin Cancer Res
2019
;25(15):
4723
34
.
22.
Song
H
,
Hedayati
M
,
Hobbs
RF
,
Shao
C
,
Bruchertseifer
F
,
Morgenstern
A
, et al
Targeting aberrant DNA double-strand break repair in triple-negative breast cancer with alpha-particle emitter radiolabeled anti-EGFR antibody
.
Mol Cancer Ther
2013
;
12
:
2043
54
.
23.
Wickstroem
K
,
Hagemann
UB
,
Cruciani
V
,
Wengner
AM
,
Kristian
A
,
Ellingsen
C
, et al
Synergistic effect of a mesothelin targeted thorium-227 conjugate in combination with DNA damage response inhibitors in ovarian cancer xenograft models
.
J Nucl Med
2019
. pii: jnumed.118.223701.
24.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
25.
Abazeed
ME
,
DJ
,
Hurov
KE
,
Tamayo
P
,
Creighton
CJ
,
Sonkin
D
, et al
Integrative radiogenomic profiling of squamous cell lung cancer
.
Cancer Res
2013
;
73
:
6289
98
.
26.
Franken
NA
,
Rodermond
HM
,
Stap
J
,
Haveman
J
,
van Bree
C
.
Clonogenic assay of cells in vitro
.
Nat Protoc
2006
;
1
:
2315
9
.
27.
Cai
Z
,
N
,
Liu
WJ
,
Chan
C
,
Pignol
JP
,
Reilly
RM
.
Optimized digital counting colonies of clonogenic assays using ImageJ software and customized macros: comparison with manual counting
.
2011
;
87
:
1135
46
.
28.
Roeske
JC
,
Stinchcomb
TG.
Dosimetric framework for therapeutic alpha-particle emitters
.
J Nucl Med
1997
;
38
:
1923
9
.
29.
Joe
H.
Relative entropy measures of multivariate dependence
.
J Am Statist Assoc
1989
;
84
:
157
64
.
30.
Linfoot
EH.
An informational measure of correlation
.
Information and Control
1957
;
1
:
85
89
.
31.
Cancer Genome Atlas Research N
.
Comprehensive genomic characterization defines human glioblastoma genes and core pathways
.
Nature
2008
;
455
:
1061
8
.
32.
Olshen
AB
,
Venkatraman
ES
,
Lucito
R
,
Wigler
M
.
Circular binary segmentation for the analysis of array-based DNA copy number data
.
Biostatistics
2004
;
5
:
557
72
.
33.
Korn
JM
,
Kuruvilla
FG
,
McCarroll
SA
,
Wysoker
A
,
Nemesh
J
,
Cawley
S
Integrated genotype calling and association analysis of SNPs, common copy number polymorphisms and rare CNVs
.
Nat Genet
2008
;
40
:
1253
60
.
34.
Mermel
CH
,
Schumacher
SE
,
Hill
B
,
Meyerson
ML
,
Beroukhim
R
,
Getz
G
.
GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers
.
Genome Biol
2011
;
12
:
R41
.
35.
Barbie
DA
,
Tamayo
P
,
Boehm
JS
,
Kim
SY
,
Moody
SE
,
Dunn
IF
, et al
Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1
.
Nature
2009
;
462
:
108
12
.
36.
Li
B
,
Dewey
CN
.
RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome
.
BMC Bioinformatics
2011
;
12
:
323
.
37.
Alford
AI
,
Kozloff
KM
,
Hankenson
KD
.
Extracellular matrix networks in bone remodeling
.
Int J Biochem Cell Biol
2015
;
65
:
20
31
.
38.
Muller
P
,
Bulnheim
U
,
Diener
A
,
Lüthen
F
,
Teller
M
,
ED
, et al
Calcium phosphate surfaces promote osteogenic differentiation of mesenchymal stem cells
.
J Cell Mol Med
2008
;
12
:
281
91
.
39.
Zhang
Z
,
Wang
J
,
Lu
X
.
An integrated study of natural hydroxyapatite-induced osteogenic differentiation of mesenchymal stem cells using transcriptomics, proteomics and microRNA analyses
.
Biomed Mater
2014
;
9
:
045005
.
40.
Itoh
K
,
Chiba
T
,
Takahashi
S
,
Ishii
T
,
Igarashi
K
,
Katoh
Y
, et al
An Nrf2/small Maf heterodimer mediates the induction of phase II detoxifying enzyme genes through antioxidant response elements
.
Biochem Biophys Res Commun
1997
;
236
:
313
22
.
41.
Itoh
K
,
Wakabayashi
N
,
Katoh
Y
,
Ishii
T
,
Igarashi
K
,
Engel
JD
, et al
Keap1 represses nuclear activation of antioxidant responsive elements by Nrf2 through binding to the amino-terminal Neh2 domain
.
Genes Dev
1999
;
13
:
76
86
.
42.
Shibata
T
,
Ohta
T
,
Tong
KI
,
Kokubu
A
,
Odogawa
R
,
Tsuta
K
, et al
Cancer related mutations in NRF2 impair its recognition by Keap1-Cul3 E3 ligase and promote malignancy
.
PNAS
2008
;
105
:
13568
73
.
43.
Cancer Cell Line Encyclopedia
.
Available from
44.
Liang
K
,
Lu
Y
,
Jin
W
,
Ang
KK
,
Milas
L
,
Fan
Z
.
Sensitization of breast cancer cells to radiation by trastuzumab
.
Mol Cancer Ther
2003
;
2
:
1113
20
.
45.
Zhang
Y
,
Kwok-Shing Ng
P
,
Kucherlapati
M
,
Chen
F
,
Liu
Y
,
Tsang
YH
, et al
A pan-cancer proteogenomic atlas of PI3K/AKT/mTOR pathway alterations
.
Cancer Cell
2017
;
31
:
820
32
.
46.
Chittenden
SJ
,
Hindorf
C
,
Parker
CC
,
Lewington
VJ
,
Pratt
BE
,
Johnson
B
, et al
A phase 1, open-label study of the biodistribution, pharmacokinetics, and dosimetry of 223Ra-dichloride in patients with hormone-refractory prostate cancer and skeletal metastases
.
J Nucl Med
2015
;
56
:
1304
9
.
47.
Ljungberg
M
,
Strand
SE
.
A Monte Carlo program for the simulation of scintillation camera characteristics
.
Comput Methods Programs Biomed
1989
;
29
:
257
72
.
48.
Pacilio
M
,
Ventroni
G
,
Cassano
B
,
Ialongo
P
,
Lorenzon
L
,
Di Castro
E
, et al
A case report of image-based dosimetry of bone metastases with alpharadin ((223)Ra-dichloride) therapy: inter-fraction variability of absorbed dose and follow-up
.
Ann Nucl Med
2016
;
30
:
163
8
.
49.
Haro
KJ
,
Scott
AC
,
Scheinberg
DA
.
Mechanisms of resistance to high and low linear energy transfer radiation in myeloid leukemia cells
.
Blood
2012
;
120
:
2087
97
.
50.
Pacilio
M
,
Ventroni
G
,
De Vincentis
G
,
Cassano
B
,
Pellegrini
R
,
Di Castro
E
, et al
Dosimetry of bone metastases in targeted radionuclide therapy with alpha-emitting (223)Ra-dichloride
.
Eur J Nucl Med Mol Imaging
2016
;
43
:
21
33
.
51.
Wu
LJ
,
Randers-Pehrson
G
,
Xu
A
,
Waldren
CA
,
Geard
CR
,
Yu
ZL
, et al
Targeted cytoplasmic irradiation with alpha particles induces mutations in mammalian cells
.
Proc Natl Acad Sci U S A
1999
;
96
:
4959
64
.
52.
Tartier
L
,
Gilchrist
S
,
Burdak-Rothkamm
S
,
Folkard
M
,
Prise
KM
.
Cytoplasmic irradiation induces mitochondrial-dependent 53BP1 protein relocalization in irradiated and bystander cells
.
Cancer Res
2007
;
67
:
5872
9
.
53.
Atkinson
MJ
,
Spanner
MT
,
Rosemann
M
,
Linzner
U
,
Müller
WA
,
Gössner
W
, et al
Intracellular sequestration of 223Ra by the iron-storage protein ferritin
.
2005
;
164
:
230
3
.
54.
Smid
M
,
Spanner
MT
,
Rosemann
M
,
Linzner
U
,
Müller
WA
,
Gössner
W
.
Subtypes of breast cancer show preferential site of relapse
.
Cancer Res
2008
;
68
:
3108
14
.
55.
Barendsen
GW
,
Koot
CJ
,
Van Kersen
GR
,
Bewley
DK
,
Field
SB
,
Parnell
CJ
.
The effect of oxygen on impairment of the proliferative capacity of human cells in culture by ionizing radiations of different LET
.
Int J Radiat Biol Relat Stud Phys Chem Med
1966
;
10
:
317
27
.
56.
Weinstein
JN
,
Collisson
EA
,
Mills
GB
,
Shaw
KR
,
Ozenberger
BA
,
Ellrott
K
, et al
The cancer genome atlas pan-cancer analysis project
.
Nat Genet
2013
;
45
:
1113
20
.