**4. Discussion**

We report the use of gold nanostars as theranostic agent againsts cancer using their photoacoustic and CT imaging capabilities to guide PTT after intratumoral injection. This strategy has clinical potential for the ablation of superficially localized tumor masses, or for aiming at complete tumor cell eradication during tumor resection surgery. The clinical potential for a similar approach was demonstrated before, using magnetic nanoparticles [31,46].

Hereby, the efficient cellular uptake of nanostars into tumor cells is important for later in vivo applications. A first criterion is the size where two processes need to be considered: (1) optimal NP/cell interaction and (2) optimal tumor retention [47,48]. Nanoparticles having a hydrodynamic diameter above 100 nm are shown to have better tumor retention after intra-tumoral injection compared to smaller sized particles. In vitro studies, on the other hand, have shown that nanoparticles of around 50 nm diameter have the ideal size for cell interaction, while the cell uptake is less efficient for larger nanoparticles [47]. We have optimized nanostars to have a hydrodynamic diameter of around 100 nm in serum, which is the cutoff-value for optimal tumor retention and efficient tumor cell interaction. Second, the protrusions of the nanostars with a small curvature potentially increase cellular uptake by increasing their membrane interaction [48]. Third, the anionic particles used in this and previous studies show a much higher stability compared to previously described cationic particles and still result in high cellular accumulation due to electrostatic interactions with the cell membrane [47]. Consequently, we were able to show that nanostars were taken up reaching a gold concentration 11.28 ± 0.82 pg Au/cell after 24 h incubation, indicating that they are suitable for efficient passive labeling of tumor cells.

Given their favorable properties for tumor cell uptake, we demonstrated that nanostars can be used as an in vitro and in vivo theranostic agent combining both dual-modality imaging (PAI and CT) and PTT. Pure nanostar suspensions resulted in an almost equally high CNR for both imaging modalities. This is different for in vitro cell experiments, where a higher CNR for PAI is measured compared to CT. This is due to the lower background signal of PAI compared to CT. Effective PTT was possible for nanostar labeled tumor cells in vitro. In contrast, unlabeled cells remained unaffected by laser treatment. For comparing the PTT effectiveness of different nanoparticles, the irradiation power that is needed to cause cell death is used as an indicator. The radiation power of 7 W/cm<sup>2</sup> used in our study is in the same range or lower as in similar experiments that were performed with nanorods (10 W/cm2) [49], nanoshells (35 W/cm2) [50], and nanocages (5 W/cm2) [51]. The relatively low radiation power used for our nanostars indicates their efficiency when compared with many nanostars reported in the literature (15 W/cm2, 38 W/cm<sup>2</sup> and 0.2 W/cm2) [13,19,21]. The maximum absorption of the nanostars used in our study is just outside the preferred window of 750–900 nm. Here, we had to find a compromise between the reduced size to enable possible delivery through leaky blood vessels and tumor cell uptake on one side and a red-shifted absorption band which would have resulted in larger nanostars [39]. In addition to inherent characteristics of the nanoparticles (absorption coefficient, size, and wavelength), other factors like cellular labeling conditions, chemical coatings, biofunctionalization of the AuNPs, cellular properties, and location of the AuNPs in the cells are equally important when comparing AuNPs regarding the irradiation power for PTT [35,48,52,53]. Consequently, there is a need for more in-depth and standardized studies for comparing different shapes of AuNPs and their theranostic potential as well as the stability of coating material at temperatures of 37 ◦C and above [35]. Although, we did not see any adverse effects on cell biology after incubation of cells with PEGylated Au-nanostars in a previous study [22], it cannot be excluded that the non-covalent Au-S-bond shows instability at elevated temperatures [35].

In addition to in vitro experiments, we evaluated the use of in vivo imaging modalities (CT and PAI) to monitor intra-tumor nanostar delivery and their corresponding contrast generating capabilities. With nanostars having optimal dimensions for intra-tumoral delivery, efficient nanostar accumulation in the tumor 24 h after injection was confirmed by PAI and CT in vivo and ex vivo TEM. These nanostars were densely packed in endosomes, as visualized by TEM. Concerning the contrast generation of these

nanostars in vivo, for PAI a significantly higher CNR (approx. 80) was observed compared to CT (approx. 25). The high sensitivity of PAI could potentially be further improved by coating the nanostars with a silica shell [54]. An additional advantage of PAI over CT is its high temporal resolution, while CT provides additional anatomical information. We were able to demonstrate that both imaging techniques can be used for intra-tumoral nanostar detection if high local concentrations can be achieved. By combining CT and PAI in a dual-modality approach using highly sensitive gold nanostars, PAI could provide rapid and detailed information on local nanostar distribution and their temporal changes, while CT could provide a full body scan with more detailed anatomical information. Compared to dual contrast agents that use radio-nuclei labeling like for PET-MRI [55,56], the combination of CT and PAI provides long-lasting contrast.

We have also confirmed the efficiency of the nanostars for PTT. Due to their star shape, they have a high absorption coefficient resulting in a large temperature increase of Δ25 ◦C when irradiated with a laser power of only 2 W/cm2. Reaching a local temperature higher than 43 ◦C will result in necrosis of tumor cells [28,57]. In vivo necrosis was confirmed in this study after tumor irradiation (2 W/cm2), while the control tumors did not show any side effects of laser irradiation. Using laser powers higher than the prescribed laser norm of 0.2 W/cm2 defined by the American Laser Institute did not damage tissue when irradiating the control sample as confirmed by histology (Figure 7) [58]. Compared to nanoshells (4 W/cm2) [50] and nanorods (2 W/cm2) [30], the nanostars require the same magnitude of laser power to induce ablation after intratumoral injection. However, comparisons between different nanoparticles to assess their therapeutic efficiency are difficult due to the different experimental conditions used and incomplete or limited information available.

The in vivo efficiency of PTT was shown in detail using BLI, MRI, and histology (Figure 6). The BLI signal intensity decrease one day after irradiation was either caused by a decrease in tumor cell viability, destroyed tissue, vessel patency or a combination of them. As a significant decrease in tumor volume is noticed by MRI from day 8 onwards, the decrease in BLI signal is most likely caused by tumor cell death. The large number of necrotic cells seen in H&E staining of the nanostar-injected tumors confirmed this hypothesis. Regrowth of the tumor after PTT treatment was observed by BLI after day 8. This is most likely due to an incomplete delivery of gold nanostars to all tumor cells so that some residual tumor tissue/cells remain as also confirmed by MRI and histology after 15 days. As local injections of nanostars will result in an inhomogeneous distribution within the tumor, applying multiple injections and irradiations will most likely improve outcome [23]. Hereby, the combination with imaging techniques that provide information on the nanostar distribution and therapy outcome is of utmost importance. Additional treatment options to improve outcome are a combination of PTT with chemotherapy, photodynamic therapy, or the introduction of tumor-targeting moieties to the nanostars [59,60].
