General surgical stereotactic systems have been known since the early 1990s and enable surgeons to virtually plan and evaluate surgical strategies prior to the actual surgery using the available imaging data from CT or MR imaging. Immobile skeletal structures such as the rigid neurocranium provided the optimal setting for precise image guidance. Therefore, initial experiences with computer-assisted and image-guided surgery were mainly gained from rigid structures and applied in neurosurgery, complex craniofacial surgery, and orthopedic surgery . In contrast, the liver parenchyma is exposed to constant breathing motion and organ deformation. Only more recently for HPB surgery have first image guidance systems taken the complexity of the liver itself into account to a sufficient level to permit their application in daily clinical practice [38, 39]. Nevertheless, bridging the gap between virtual pre-operative intervention planning and intra-operative navigation support for the real-time reproduction of the virtually planned intervention remains a challenge. Deformations of the liver of up to 20 mm have been detected when compared to a pre-operatively acquired 3D tomographic image . Thus, the fundamental challenge in image guidance in liver interventions is a correct and real-time referencing of the pre-operatively acquired image data (3D- and 4D-/contrast-enhanced/CT or MRI) with the actual intra-operative situation. Such a navigational support of an optically based instrument guidance system is already in use in our hospital for open liver surgery  and MWA . Currently, such image-guided navigation systems are not used for the placement of applicators for liver HDR-BT. In the present study, we demonstrate the potential use of this available image-guidance technology for interventional procedures with specific adaptions to accommodate the placement of HDR-BT applicators.
Initial studies of intra-operative liver brachytherapy first described the technique in the 1980s [20,21,22]. At that time, applicator placement was guided either by manual palpation or sonographic monitoring . In such a situation, neither adequate coverage of the target volume nor exact sparing of adjacent critical structures could be guaranteed. A decade ago, image-guided brachytherapy of liver tumors was introduced , in which an iridium-192 source is temporarily inserted into the tumor through catheters placed under CT image guidance [11, 23,24,25]. In the present study, we demonstrate an effective geometric accuracy of 4.2 mm ± 1.2 mm for the placement of the HDR-BT applicators using such an available image-guidance technology for interventional procedures, with a potential degrading of V100% in the PTV from 97 to 87% (Fig. 7). In a subsequent study at our department , we analyzed the dosimetric impact of geometric uncertainties (i.e., tumor size, shape, number of lesions, needle length and position, number of needles as well as OAR and target volume coverage) in navigated HDR-BT for liver tumors by applying different shifts (≤5 mm and ≤1 cm) to a known dose distribution. It was found that for small PTVs below 2 cm, the deviations for dose coverage and dose conformity can be up to 50%. The effect depends on the number as well as on the placement and arrangement of needles and decreases with increasing tumor size. For larger PTVs, the reduction in dose coverage for a ≤5 mm shift is between 5 and 10%. It was further concluded that where no other treatment modality is available, such a navigation-guidance method for intra-operative HDR-BT seems to be promising. However, specific inclusion/exclusion criteria for patient selection for navigated intra-operative HDR-BT remain unclear at this stage.
In clinical practice, several possibilities exist to overcome these uncertainties. Using more than one applicator for a liver brachytherapy plan enables better coverage of complex-shaped tumor volumes, as well as better protection of the healthy surrounding tissue with its rapid dose fall-off outside the target lesion. It also increases the robustness of the plan. Additionally, the non-optimal placement of the brachytherapy needle can be compensated by a corresponding adjustment of the dwell time of the source within the target volume. With these various adjustable elements for planning, the minimum target dose at the tumor border can be achieved with no compromise in target coverage. Since the goal in liver brachytherapy is to cause necrosis in the tumor, dose heterogeneity within the target volume is of no concern and no maximum dose limit exists [27, 32].
The assessment of the present datasets indicates the feasibility of navigated intra-operative HDR-BT. The method is potentially feasible for all these patients with either a hepatic tumor load below 40% of the total liver volume or less than six liver metastases larger than 4 cm. For the selected datasets, acceptable doses to the normal liver as well as acceptable treatment times have been observed. The risk of harming blood vessels by placing applicators in liver brachytherapy is potentially mitigated through the use of the described guidance method by reducing the need for re-placement due to inaccuracy.
In clinical practice we often face complex settings in patients with CRC hepatic metastases, such as multiple, synchronous, bi-lobar, and large hepatic lesions that are located near critical structures or blood vessels. Combining this navigation-guided intraoperative HDR-BT with surgery and MWA also opens new opportunities for a combined treatment approach at the same time. Such a computer-navigated system may facilitate the application of HDR-BT for the placement of BT applicators and planning of liver HDR-BT also in situations where HDR-BT was not initially planned to be part of the intervention.
With regards to the dose prescription, in the present study we prescribed 20 Gy to at least 90% of the PTV. Previous studies of intraoperative brachytherapy for liver cancers proposed that tumor-enclosing isodoses of 15 to 30 Gy should be used [20,21,22, 28]. In a prospective randomized trial of 73 patients with 199 colorectal liver metastases, Ricke et al.  evaluated the minimal dose levels enclosing the CTV of 15, 20, and 25 Gy. They found a dose dependency for local control if the minimum dose in the target exceeded 20.4 Gy. In a multivariate analysis, lesion size (range of tumors 1–13.5 cm) had no impact on local control. However, in tumors larger than 8 cm, excessive irradiation time and unpredictable risks aggravated the delivery of >20 Gy in a single fraction. Nevertheless, these data show that lesions up to a diameter of 8 cm can be safely treated with a single fraction of 20 Gy applied by HDR-BT.
There are several limitations to this feasibility study. First, any conclusions revealed here are hypothesis-generating and, as such, will need to be validated within a prospective study. Several open questions for intraoperative navigation-guided HDR-BT remain: it is unclear how many applicators need to be placed for lesions > 8 cm and what conformity should be achieved, especially in bigger lesions. With only one applicator, dose conformity might be too low. More applicators might provide a better conformity of dose distribution. We suggest the number of applicators should depend on the size of the lesion. Also, dose constraints for liver HDR-BT are unclear. In their studies, Ricke et al.  used a maximum dose exposure of 5 Gy to maximally two thirds of the liver volume as a liver constraint. In their 2005 published study  on dose–response relationship for small volumes of liver parenchyma after single-fraction HDR irradiation, they used, after brachytherapy, a hepatocyte-specific contrast agent media uptake (Gd-BOPTA) for liver MRI as a surrogate for liver function. They found that the lowest threshold dose to impair hepatocyte function was 9.9 Gy (standard deviation 2.3 Gy) at 6 weeks after irradiation with repair ongoing until 3 months. After single-fraction stereotactic body radiotherapy (SBRT) to liver malignancies, Herfarth et al.  observed a focal reaction after contrast CT at a dose minimum of 13.7 Gy (range 8.9–19.2), which is consistent with the experience in brachytherapy. In the current study, we have used the dose constraints for stereotactic body radiation therapy (SBRT) according to Grimm et al. , with which technique also high doses of single fractions can be applied .
Since the technology of this system was initially developed for intra-operative support of liver resection and MWA, we have used this clinical setting to evaluate the potential benefit of the use of such a navigation system for liver BT. However, the use of this stereotactic navigation system opens also new opportunities for the placement of HDR-BT applicators within the liver by percutaneous CT guidance in patients with large, centrally localized liver metastases where no resection or MWA intervention is indicated. It may provide support in pre-interventional treatment planning (i.e., needle path, multi-applicator configuration) and applicator placement (i.e., angle and depth guidance).
In conclusion, we provide evidence for the technical feasibility of navigation-guided intra-operative HDR-BT and propose a set of inclusion criteria for clinical testing.