Soft Tissue Sarcomas (STS) represent a diverse category of malignancies exhibiting significant clinical and pathological variation. Collectively, these tumors comprise under 1% of adult cancers but account for approximately 15% of pediatric malignancies.¹ Diagnosis often occurs when tumors are large, with the extremities being the most common primary site.^2–3 Clinical evidence supports the National Comprehensive Cancer Network (NCCN) recommendation for either preoperative or postoperative radiotherapy (RT) in large (> 8cm), high-grade STS. The standard neoadjuvant external beam RT (EBRT) regimen is 50 Gy, delivered in fractions of 1.8 to 2.0 Gy.¹ The advancement of radiation therapy technologies has established intensity-modulated radiation therapy (IMRT) combined with image guidance radiotherapy (IGRT) as the preferred EBRT modality, primarily due to its superior ability to minimize toxic effects.⁴ Despite their implementation being linked to reduced toxicity, IMRT combined with IGRT has not demonstrated superiority in survival or disease control outcomes.⁵

GRID radiation therapy technique was originally developed by Alban Köhler in the 1950s for safe treatment of bulky tumors using kV x-ray technology.^6–7 In the megavoltage radiotherapy era, GRID has been repurposed for palliative management of large tumors with positive outcomes, now formally termed spatially fractionated radiotherapy (SFRT).^8–15 Emerging in 2010 as a three-dimensional extension, LATTICE radiotherapy (LRT) leverages contemporary systems (IMRT, volumetric modulated arc therapy (VMAT), robotic convergent beams, charged particle Bragg peaks) to create focal high-dose vertices within tumors. This generates characteristic peak-valley dose distributions while sparing surrounding tissues.^16–17 Owing to its superior normal tissue sparing compared to GRID, LRT represents a rational strategy for further dose escalation.

The implementation of LRT is feasible through multiple platforms including MLC-based linacs, Cone-based robotic radiosurgery (CyberKnife), and particle therapy units (proton/carbon ions), though each approach entails unique trade-offs in clinical application.^17–19 Dosimetric analyses of photon, proton, and carbon-ion LRT modalities revealed comparable peak-to-valley dose ratios (PVDR). Nevertheless, particle-based approaches (proton/carbon-ion) demonstrated superior organ-at-risk (OAR) sparing relative to photon techniques.¹⁹ Gamma Knife's status as the intracranial SRS gold standard stems from its ability to create sharp dose fall-offs at small target boundaries, sparing critical organs.^20–21 Such dosimetric characteristics may enable higher PVDR values when implementing LRT techniques.

A novel TaiChi radiotherapy platform, featuring an integrated linear accelerator (LINAC), focused gamma system, and kV imaging subsystem within a sealed O-ring gantry, has entered clinical use(NMPA: 20223050973; FDA: K210921).The platform facilitates synchronous administration of photon-based IMRT and gamma knife-equivalent SRS, consolidating dual modalities in a single treatment ecosystem. This consolidated platform offers potential benefits for patients requiring sequential LRT followed by conventional fractionation external beam radiation therapy (cEBRT):The linac component delivers homogeneous dose distributions to the planning target volume (PTV), while the gamma-knife’s submillimeter accuracy enables focused higher dose escalation to LRT vertices. This study investigates Taichi’s dosimetric advantages in LRT for large STS, hypothesizing that it outperforms conventional VMAT in vertex dose escalation, dose gradient quality, and OAR protection.

Ten patients with histologically confirmed large soft tissue sarcomas of the lower extremity (Tumor size ≥ 8 cm in maximum diameter) were retrospectively selected from our institution's digital medical archives. Patients were immobilized using a vacuum-based fixation system to minimize intrafraction motion.Simulation was performed using the Philips Brilliance Big Bore CT scanner with a 1 mm slice thickness. Gross tumor volume (GTV) was delineated on planning CT using MRI fusion, while clinical target volume (CTV) encompassed a 1.5 cm margin from the GTV in radial direction and 4 cm in superior/inferior direction, and the CTV was uniformly expanded by 5 mm in all three dimensions to generate the PTV. Within the GTV, 0.9–1.5 cm spherical high-dose vertices were geometrically configured with 3.0-3.5 cm center-to-center separation. These vertices occupied 1–10% of GTV volume (modal value: 2%).¹⁷ Nerve and bone were constrained as OARs. An inward LATTICE margin (1-2cm) maintained vertex-to-boundary distance to spare normal tissues.

The treatment consisted of two courses: 1) LRT: 15 Gy single fraction to high-dose vertices; 2) cEBRT: 50 Gy in 25 fractions (2 Gy/fx) to PTV. LRT prescription necessitates defining three dosimetric parameters: peak dose (covering ≥ 95% of high-dose vertices), valley dose (minimum dose within GTV), and tumor peripheral dose (maximum allowable dose at tumor margin, typically ≤ 5 Gy) to mitigate normal tissue toxicity.

a)Linac Planning

Clinical treatment plans were developed on a Varian iX linac (6MV) using Eclipse 13.5 TPS with VMAT optimization. For LRT delivery, six co-planar partial arcs (220°rotation) avoided contralateral leg irradiation, employing collimator angles of 45°and 315°to deliver 15 Gy in single fraction to high-dose vertices. The cEBRT phase utilized two partial VMAT arcs for conventional fractionation (50 Gy total, 2 Gy/fx) to PTV. All dose calculations employed the anisotropic analytical algorithm (AAA) at 2 mm grid resolution. Patient-specific quality assurance (QA) for LRT plans implemented portal dosimetry with a 2%/2 mm and a 10% dose threshold criterion.

b)TaiChi Planning

The TaiChi treatment plans were implemented in RTPRO TPS (OUR United Corp., China) using dual-modality optimization strategy: Gamma knife-based LRT plan was initially optimized to cover the high-dose vertices (15 Gy/1 fx). Subsequent 6MV linac VMAT optimization achieved PTV coverage (50 Gy/25 fx) with parameters consistent with conventional protocols described previously. Dose calculations employed CCC algorithm at 2 mm resolution. Patient-specific QA for LRT utilized IBA myQA SRS systems with a 2%/2 mm and a 10% dose threshold criterion.

c)Comparison metrics

Dosimetric comparisons between the clinical Linac plans and Taichi plans were made to evaluate the ability of the technique to deliver high doses to the vertices while minimizing dose to OARs and the non-LATTICE target volume within the GTV. Dose heterogeneity was quantified via the peak/valley dose ratio (PVDR). In accordance with prior studies ^26–28, the conventional PVDR metric was replaced by the D₁₀/D₉₀ ratio, defined as the dose covering 10% of the GTV volume (D₁₀) divided by that covering 90% (D₉₀).Target dose analysis included the comparison of D₉₅ (dose covering 95% volume), D_0.5cc (dose covering 0.5cc volume), and mean dose (D_mean) for vertices. The dosimetric evaluation of cEBRT plans was performed using the dose conformity index (CI) and homogeneity index (HI). The CI, defined as (PTV_Dp/PTV)×(PTV_Dp/V_Dp)-where PTV_Dp is the planning target volume receiving the prescription dose and V_Dp is the total volume enclosed by the prescription isodose line།ranges from 0 to 1, with values closer to 1 indicating better conformity. The HI, calculated as (D₂།D₉₈)/D₅₀ (where D₂, D₅₀, and D₉₈ represent the doses covering 2%, 50%, and 98% of the target volume, respectively), reflects dose homogeneity, with lower values indicating a more homogeneous dose distribution within the target. The OARs evaluation encompassed D_max and D_mean for nerve, bone. Statistical analysis employed two-tailed paired t-tests with a significance threshold of p < 0.05.

The GTVs ranged from 208.8 to 627.4 cm³, with a mean GTV of 419.3 cm³. The number of high-dose vertices ranged from 7 to 15, and the vertices volume accounted for 1.9% to 2.8% of the GTV (mean 2.3%).

All treatment plans met clinical dosimetric criteria. This study focuses on evaluating the comparative performance of LRT plans delivered via conventional linac versus gamma knife systems. Consequently, our analysis emphasizes the distinctions between these two LRT modalities. Representative dose distributions and dose-volume histograms (DVHs) for TaiChi and conventional linac plans are displayed in Fig. 2 (one selected case). Comparative dosimetric parameters for targets and OARs of LRT plans are showed in Table 1 and Fig. 3. The TaiChi plans demonstrated significantly enhanced dose delivery to vertices compared to conventional linac plans, with higher D_mean (18.87 ± 0.46Gy vs. 16.66 ± 0.85Gy, p < 0.0001) and D_0.5cc(25.15 ± 0.39Gy vs. 18.90 ± 0.44Gy, p < 0.0001). Furthermore, The TaiChi system demonstrated significantly improved dose gradient characteristics, with higher higher D₁₀/D₉₀ values of GTV(5.93 ± 0.84Gy vs. 3.40 ± 0.92Gy, p < 0.0001) and lower doses of GTV margin (3.95 ± 0.48Gy vs. 4.87 ± 0.41, p < 0.0001). The OAR dose metrics are listed in Table 1. The TaiChi plans achieved comparable or reduced OAR exposure, with statistically significant reductions in D_max for nerve, bone (all p < 0.05). Taichi improved vertex dose escalation by 33% and OAR sparing by 15–35%. The Taichi’s γ-ray component enabled sharper dose gradients, reducing non-target tumor irradiation. Regarding cEBRT plans, both approaches utilize linac-based VMAT techniques. Comparative dosimetric parameters for targets and OARs of cEBRT plans are showed in Table 2 and Fig. 4. No statistically significant differences were observed in either targets or OARs, consequently, the cEBRT plans are not discussed in detail in this study.

All LRT plans demonstrated clinically acceptable patient-specific QA results, achieving gamma-index passing rates exceeding 95% with a stringent 2%/2mm (10% low dose threshold) criterion. Specifically, linac plans demonstrated a pass rate of 97.2% ± 1.3% using portal dosimetry verification, while TaiChi plans achieved 97.1% ± 1.0% via IBA myQA SRS system validation for patient-specific QA. Figure 5 demonstrates the Patient-specific QA results of a representative TaiChi LRT plan, achieving a 97.5% gamma passing rate with 2%/2mm (10% low dose threshold) criteria using an IBA myQA SRS system.

Figure 5. Patient-specific QA results of a representative TaiChi LRT plan using an IBA myQA SRS system. Left: plan dose distribution; Middle: measured dose distribution; Right: gamma index image with 2%/2mm (10% low dose threshold), passing rate 97.5% .

In this study, we conducted a comparative analysis of the clinical dosimetry advantages offered by bimodal X-ray/γ-ray LRT for soft tissue sarcomas using the Taichi platform. To the best of our knowledge, this represents the first systematic investigation evaluating the dosimetric benefits between the innovative Taichi platform and conventional linear accelerators in LRT applications. Our findings demonstrate that this integrated approach facilitates precise high-dose radiation delivery to target vertices while preserving sharp dose gradients and sparing adjacent critical tissues. Previous research has revealed that the accelerator component (X-ray) of the Taichi platform achieved comparable dosimetric outcomes to traditional C-arm linacs in cervical cancer treatment planning, while demonstrating significantly superior performance in breast cancer cases.²⁹ Additionally, multiple studies have demonstrated that the TaiChi dual-modality system (X-ray/γ-ray) provides superior dose escalation outcomes for both prostate and pancreatic cancers, achieving improved target conformity while reducing radiation dose to adjacent OARs compared to conventional linear accelerators. ³⁰

SFRT encompasses multiple modalities including GRID, LATTICE, micro-beam, and mini-beam techniques.^31–35 As a recently emerged novel approach, preliminary clinical data demonstrate LRT's favorable safety profile with no severe treatment-related toxicities reported to date. Tumor response rates mirror those observed in GRID therapy, confirming technical feasibility.^33,36 The delivery of ablative radiation to selective tumor subvolumes achieves significant debulking, with additional localized dose escalation (boost therapy) to viable tumor regions resulting in substantially improved tumor control rates when normal tissue dose limits are respected. Maximizing peak-to-valley dose differentials remains clinically advantageous, but current photon technologies exhibit inherent limitations in achieving optimal valley dose reduction.¹⁷

As the first implementation of Gamma Knife in LRT, our study demonstrates distinct dosimetric advantages: higher PVDR(D₁₀/D₉₀) combined with reduced peripheral doses at GTV margins and diminished exposure to OARs. These dosimetric advantages originate from three fundamental characteristics of Taichi’s Gamma Knife technology: (1) superior penumbral sharpness inherent to gamma rays compared to X-ray modalities; (2) an intrinsic spherical dose-painting mechanism—physically congruent with spatially fractionated radiotherapy principles—that generates higher PVDR than MLC-based linac systems; and (3) dynamic rotational isocentric delivery enabled by hemispheric convergence of 18 Co-60 sources synchronized with continuous gantry rotation, collectively steepening target dose gradients through multiplanar focusing. Furthermore, the TaiChi platform pioneers the integration of kV X-ray and γ-ray modalities within a single unit. As established earlier, its Gamma Knife component delivers LRT to high-dose vertices (e.g., 15 Gy ×1), while the linac-based X-ray system administers cEBRT to the PTV (e.g., 2 Gy ×25). This dual-capability design precisely aligns with the combined LRT + cEBRT dose paradigm recommended for soft tissue sarcomas—establishing TaiChi as an optimized radiotherapy platform for implementing this therapeutic strategy.

This study presents several limitations. First, the modest cohort size (n = 10) and retrospective nature constrain the generalizability of clinical inferences. Prospective randomized trials incorporating larger patient populations are required to establish robust correlations between observed dosimetric benefits and clinically meaningful endpoints including progression-free survival, locoregional control, and late toxicity profiles. Second, critical parameters for LRT vertex configuration - including optimal spatial distribution and geometric characteristics - remain to be standardized. Future studies employing computational radiomics or hypoxia-specific PET biomarkers may facilitate biologically-guided vertex positioning to target radioresistant tumor subvolumes.

The TaiChi platform's innovative integration of LINAC and Gamma Knife technologies provides distinct dosimetric advantages for STS LRT, enabling superior dose escalation to high-dose vertex regions while maintaining steep dose gradients and effective normal tissue sparing. Physical advantages include: 1) γ-ray focusing for precise vertex dose escalation, 2) LINAC-based modulation for conventional target coverage, and 3) synergistic dose distribution optimization. These capabilities position TaiChi as a promising platform for advancing LRT applications. Further clinical validation and protocol optimization are warranted to translate these physical advantages into improved patient outcomes.