Preparation of PIDA and ultrahigh X-ray attenuation
The formation of PIDA requires the topochemical polymerization, i.e., the head-to-tail 1,4-polymerization, of diiodobutadiyne monomers assembled in an appropriate arrangement (Fig. 1a). To achieve the required spacial alignment of monomers for this topochemical polymerization, we use host-guest supramolecular scaffolds based on halogen bonds, the Lewis acid-base interactions between the monomer and functionalized Lewis bases44, 46. In previous attempts to prepare PIDA42, 43, 47, biscyanoalkyl or bispyridyl oxalamides were employed as hosts to align monomer guests in co-crystals. However, they could not yield PIDA with refined structure in large quantities43, or only generated co-crystals that required high pressure to initiate the polymerization48. Here, using N1,N2-bis(2-(pyridin-3-yl)ethyl)oxalamide (1, structure shown in Supplementary Fig. 1) as the host49, we are able to produce PIDA co-crystals with well-defined structure (Fig. 1b) on gram-scale, so that the possibility of further application of PIDA is boosted. PIDA•1 co-crystals exhibit a metallic appearance that is characteristic of a high degree of polymerization (Fig. 1b, bottom right), and scanning electron microscope (SEM) image reveals their very smooth surface, with one dimension (010) much smaller than the other two dimensions (Fig. 1b, top right).
In order to isolate PIDA from the co-crystals and formulate it in aqueous media for in vivo applications, we utilized an amphiphilic polymer, polyethylene glycol-grafted poly(maleic anhydride-alt-1-octadecene), or C18-PMH-PEG, as the surfactant50. Sonication followed by dialysis of a mixture of the PIDA•1 co-crystals and C18-PMH-PEG in water resulted in a stable blue aqueous suspension (Fig. 1c). The UV-visible absorption spectrum of the blue PIDA suspension (Fig. 1d) exhibited a maximal absorption peak at 652 nm with a broad shoulder at 750 nm, attributed to the planar backbone of the material. The Raman spectrum of the suspension was consistent with that of the co-crystals (Fig. 1e), indicating that the structure of PIDA was preserved during the process. Transmission electron microscope (TEM) imaging confirmed that the blue suspension mainly contained well-dispersed nanofibers with diameters of 30–50 nm (Fig. 1f). Moreover, the elemental analysis of the nanofibers by an energy-dispersive X-ray (EDX) detector showed the existence of only carbon (15.7 wt%) and iodine (84.3 wt%) atoms, which was consistent with the theoretical value of PIDA (15.9 wt% for carbon and 84.1 wt% for iodine) and evidenced the thorough removal of host 1 (Fig. 1g).
Effective payload of iodine atoms is crucial to the high performance of CT contrast agents12. However, current clinically available CT contrast agents all have iodine mass contents of less than 50%, such as iohexol (46.4 wt%), iopromide (48.1 wt%), and iodixanol (49.1 wt%). The 84.1 wt% iodine content of PIDA warranted it an ultrahigh X-ray attenuation ability. The CT value, an indicator in terms of Hounsfield unit (HU) that measures the ability of a material to attenuate X-rays, of the PIDA suspension increased linearly with the iodine concentration (Fig. 1h), indicating that PIDA was well dispersed in the media. The slope of HU values against sample concentrations for PIDA was 1.8 times higher than that for iohexol, suggesting a superior X-ray attenuation ability of PIDA. More strikingly, when a PIDA suspension was condensed to form PIDA aggregates (Supplementary Fig. 2), its CT value dramatically increased over 10 times, from 213 HU to 2475 HU (Fig. 1i). To validate this aggregation-induced CT intensity amplification, we injected a PIDA suspension and an iohexol solution (both with an iodine concentration of 4 mg mL−1) into two porcine tissues, respectively (Supplementary Fig. 3). The contrast ratio of PIDA in the pork muscle tissue (87.8%) was 4 times higher than that of iohexol (21.7%), whereas in the fat tissue, the contrast ratio of PIDA (251.7%) was 17 times higher than that of iohexol (14.5%). In addition, PIDA nanofibers were observed to accumulate in the injection sites, with a significantly longer in situ retention time than iohexol.
PIDA-based ultraefficient CT imaging in rats
Inspired by the remarkable X-ray attenuation ability of PIDA, we injected PIDA suspensions ([I]: 4 mg mL−1) into rat legs to evaluate their performance in CT imaging. Iohexol solutions ([I]: 4 mg mL−1) were injected in the control rat. Ideally, a contrast agent should improve the absolute CT attenuation of the target tissue to more than twice of the surrounding tissue and fluids12. The background CT signal of rat muscle tissue was about 50 HU, and the CT value of intramuscularly injected PIDA could reach 260 HU at 2 min after injection (Fig. 2a). As a comparison, the CT value of iohexol was around 150 HU at that time (Fig. 2b). Markedly, the CT value of PIDA continued to increase as time progressed and reached 660 HU at 2 h after injection, possibly attributed to the aggregation of PIDA at the injection site. The PIDA converged at the injection site and formed a well-defined marker with fixed geometry and position. Its CT value maintained at a high level of above 500 HU for over 6 h (Fig. 2d and Supplementary Fig. 4), a typical period required for CT-guided preoperative planning, surgery scheduling, and surgical procedure in hospital. As a comparison, the injected iohexol was barely visible at 2 min after injection, and its CT value dropped promptly to less than 100 HU within 1 h (Fig. 2b, 2d and Supplementary Fig. 5). Since an effective diagnostic dose of iohexol was typically in molar concentrations, we next applied a high concentration iohexol solution ([I]: 100 mg mL−1) in rats. Although the initially formed marker had a high CT value (1700 HU), it decayed very fast and the CT value decreased dramatically down to 100 HU at 2 h post injection (Fig. 2c, 2d and Supplementary Fig. 6), suggesting its rapid diffusion and clearance. The PIDA marker was also clearly visible under regular X-ray examination, which further confirmed the ultraefficient X-ray attenuation of PIDA (Supplementary Fig. 7). The great performance of PIDA in rat CT imaging hence demonstrates its great potential to meet clinical marker needs, given its strong CT intensity at ultralow iodine concentrations, prolonged tissue retention time, as well as high geometrical and positional stability.
CT and naked eyes dual-visible surgical marker
The practicability of current imaging markers for surgical guidance is seriously restricted in that they cannot accurately translate diagnostic image findings into specific therapeutic intervention. A surgical marker that can bridge this gap by identifying occult lesions both under instrumental guidance and by direct visual observation is highly desirable in clinic18, 51. In addition to the ultraefficient CT imaging characteristic, PIDA can preserve its unique deep blue color in vivo for various time after the injection (Supplementary Fig. 8), hence enabling precise localization of the targeting lesions by both CT imaging and naked eyes. To evaluate if PIDA can serve as such a dual-visible surgical marker, we injected PIDA suspensions on the tumor periphery of an orthotopic xenograft rat model under CT imaging guidance (Fig. 3a), and examined the assistance of PIDA labeling in identifying tumor resection margin (Fig. 3b). As expected, the injected PIDA markers maintained high level CT values with fixed geometry and positions on the tumor periphery for a period of 6 h (Fig. 3c and Supplementary Fig. 9). When we dissected the rats at 24 h after injection, the blue colored PIDA markers were still clearly visible by naked eyes and distributed around the nonpalpable tumor lesion to indicate the surgical margin of the targeted tumor tissue (Fig. 3d). As a comparison, the CT signals of iohexol control (at the same iodine dose) disappeared quickly (Supplementary Fig. 10). more strikingly, after we intratumorally injected the PIDA suspension, the PIDA nanofibers diffused and fulfilled the whole tumor tissue in 2 h (Fig. 3e), clearly demonstrating the margin line between the tumor tissue and the normal tissue. The complete tumor infiltration of PDDA nanofibers had been confirmed by magnetic resonance (MR) imaging (Fig. 3g) and CT imaging (Fig. 3h), which could be directly used to identify the tumor boundary by naked eyes in the surgery (Fig. 3f, 3i). Intraoperative identification of tumor boundary remains a great challenge in clinic, and it is especially difficult for small and nonpalpable tumor lesions. The dual-visible PIDA allowed precise intraoperative localization of occult orthotopic tumors, demonstrating a great potential to directly translate diagnosis accuracy to therapeutic intervention for enhanced clinical surgical outcomes.
Fiducial marker for stereotactic body radiation therapy
In addition to the surgical marker, PIDA can also serve as an excellent candidate for fiducial makers in stereotactic body radiation therapy (SBRT). SBRT is an advanced approach of IGRT to precision radiotherapy, which employs robotic radiotherapy (RRT) to deliver precise doses of radiation with extreme accuracy (Fig. 4a)52, 53. Because of its high precision, SBRT has to take real-time tumor movement into account, while tumor motions during the respiratory cycle is a major contributor to targeting uncertainty. RRT utilizes an orthogonal X-ray imaging system and an optical respiratory tracking system to build a real-time correlation model between the tumor and the external abdominal surface of the patient (Supplementary Fig. 11). The established correlation model then allows the RRT to estimate tumor position based on optical tracking, which is sent to the robotic positioning system to enable real-time adjustment of the tumor position during respiratory cycles. In clinic, Au-based fiducial markers are most commonly implanted in tumor tissues to serve as a surrogate for tumor positioning in clinical practice. However, they are prone to cause local edema, inflammation, and positioning shifts31–33, 52, therefore impeding patients from receiving desired clinical benefits. PIDA, on the other hand, represents a straightforward solution to the problems associated with Au-based fiducial markers.
To make a direct comparison with the commercially available Au marker, PIDA was casted into a similar shape and implanted into the liver of a rat through puncture needles under the guidance of CT imaging (Fig. 4b). On Day 7 post implantation, MR imaging revealed severe local edema caused by the implanted Au marker (Fig. 4c), which was confirmed by the optical image of the dissected tissue (yellow arrows, Fig. 4d). As a comparison, no PIDA-caused edema was observed either on MR image (Fig. 4e) or dissected tissue (Fig. 4f). The blood analysis, the hematoxylin and eosin (H&E), and TdT-mediated dUTP nick end labeling (TUNEL) stain of the tissue around the implanted markers also indicated excellent biocompatibility of PIDA as the implantable marker (Supplementary Table 1 and Supplementary Fig. 12). Moreover, the implanted Au marker contacted loosely to the surrounding tissues (Fig. 4d), which implied a high probability of position shift or even off target during clinical application. As a comparison, PIDA marker attached tightly to the surrounding tissues (Fig. 4f), suggesting a stable positioning in SBRT.
To further assess the positioning accuracy of PIDA as a fiducial marker in SBRT, we implanted a PIDA maker into the liver of a Beagle, on which a clinical SBRT was executed afterwards (Fig. 4g). A vacuum cushion was used to fix the posture of the Beagle during the SBRT treatment and to ensure the consistency with the preoperative CT modeling (Fig. 4g, bottom inset). From CT imaging, the PIDA marker exhibited a strong CT signal in the liver of the Beagle (Fig. 4g, top inset), on basis of which a three-dimensional distribution map of planned X-ray irradiation was generated for RRT on the Beagle (Fig. 4h). The real-time position of the PIDA marker in RRT was monitored by live X-ray imaging (green circle in Fig. 4i), which moved close to the original position (yellow diamond in Fig. 4i, determined from the preoperative CT imaging) periodically during the respiratory cycle of the Beagle (Supplementary Video 1). In addition, the correlation error for RRT was measured based on the difference between the actual position of the target (green circle in Fig. 4i) and the predicted position of the target computed from the Beagle’s respiratory movement (Fig. 4j). The average correlation error of the PIDA marker was 1.07±0.55 mm (Fig. 4k), lower than that of Au markers typically found in clinic (1.7±1.1 mm)54, indicating a good agreement with the correlation models for the execution of the planned RRT (Fig. 4h). The uncertainty of the PIDA marker, which provided the detection uncertainty value for the fiducial extraction algorithm, was 9.10±2.30% (Fig. 4l), far below the uncertainty threshold parameter of 40%55. PIDA hence exhibited an exceptional positioning accuracy in SBRT of the Beagle.
An effective SBRT replies not only on the accurate positioning of fiducial markers in RRT, but also on the precise radiation dose calculation, which can be easily affected by artifacts in CT images. Compared with the nontrivial metal artifacts that were seen in the CT images of the rat implanted with an Au fiducial marker (Fig. 5a), PIDA presented a much more efficient labeling effect with minimal artifact interference (Fig. 5b). To further quantify the advantage of PIDA fiducial marker in artifact reduction, we used a thorax phantom with movable plugs to simulate the CT imaging of a whole human body in SBRT (Supplementary Fig. 13). Each movable plug in the phantom simulated a specific human organ, and inserting PIDA or Au marker to a plug mimicked the labeling of the corresponding organ. The CT imaging showed that PIDA could mark the target organ much more effectively than Au, given its negligible artifacts (Fig. 5c). The quantified CT intensities on the circled spots distributing near the PIDA marker (Fig. 5d, 5e) were almost identical to those of the unmarked sample (Control), suggesting that the artifact-free PIDA marker showed no influence on the visibility of the surrounding tissues. As a comparison, the Au marker exhibited a significant decrease of CT intensity in the tangential direction of the plug, as well as a considerable increase of CT intensity in the perpendicular normal direction. In addition, the calculated radiation dose distribution for RRT planning based on the Au marker apparently deviated from the actual dose distribution in the Control group (Fig. 5f, 5g). As a comparison, the PIDA marker resulted in a more accurate dose distribution calculation (Fig. 5g), showing a significant advantage over Au-based fiducial marker in assisting the treatment planning system for the precise radiation delivery in SBRT.
Biocompatibility and Biodegradability of PIDA
To further ensure the safety of PIDA for future applications as a CT contrast agent, we comprehensively characterized its biocompatibility and biodegradability. We first mixed different concentrations of PIDA suspensions with rat red blood cells (RBCs). Triton was used as a positive control and set as 100% hemolysis. The hemolysis rates of all PIDA suspensions were below 5%, suggesting that PIDA did not cause hemolysis at tested concentrations. In addition, we incubated PIDA suspensions at different concentrations with 4T1 cells (mouse breast cancer cells), NIH 3T3 cells (mouse embryonic fibroblasts), and HEK 293T (human embryonic kidney cells) for 12 h (Fig. 6b). The cell viabilities of all PIDA-containing groups, tested by MTT assays, maintained above 90%, showing the good biocompatibility of PIDA.
Moreover, the weight of the experimental rats increased normally over time (Fig. 6c), and no significant difference in blood analysis and liver/kidney function tests between PIDA-injected rats and normal rats was observed (Supplementary Fig. 14), indicating that PIDA did not affect the physiological status of the animals. In addition, we have showed that the CT value of injected PIDA suspension ([I]: 4 mg mL−1) could maintain over 500 HU for 6 h (Fig. 2d). In fact, it still remained high at around 400 HU for 24 h (Fig. 6d, 6f), which should guarantee the completion of most CT-involved operations in hospital. The injected PIDA became invisible by CT imaging on Day 7 post injection (Fig. 6e), and on Day 21, the CT value decreased to a level close to iohexol ([I]: 4 mg mL−1), which was injected into another rat synchronously with PIDA (Fig. 6f). The CT value and the volume of the PIDA fiducial marker in rat liver also decreased over time (Supplementary Fig. 15). The gradual degradation and bioabsorbability of PIDA can be explained by the fact that polydiacetylene backbone is cleavable by reactive oxygen species56, and the relatively labile carbon-iodine bond can also facilitate the process57. In addition, no obvious damage was found in major organs at 1 day or 1 week after the injection (Supplementary Fig. 16), and the H&E stains of main organ tissues also confirmed that PIDA did not cause any toxicity during the whole process (Fig. 6g). The excellent biosafety and biodegradability of PIDA therefore pave the way for translational studies of PIDA-based CT contrast agents.