C–X–C Motif Chemokine Receptor 4 (CXCR4)‐Targeted Biologically Guided Proton Spatially Fractionated Radiotherapy for Bulky Limited‐Stage Small Cell Lung Cancer Authors: Keru Pu, Tengxiang Li, Shengnan Xu, Xue Sha, Qingtao Qiu, Jinming Yu, Yong Yin, Jie Liu Published: 2026 ABSTRACT AbstractBackground and PurposeTo evaluate the feasibility and dosimetric safety of integrating C–X–C motif chemokine receptor 4 (CXCR4)‐targeted tracer 1 8F‐AlF‐NOTA‐QHY‐04 positron emission tomography imaging with biologically guided proton spatially fractionated radiotherapy (BG‐pSFRT) for bulky limited‐stage small cell lung cancer.Materials and MethodsAmong 30 patients with bulky limited‐stage small‐cell lung cancer, a threshold range of 35%–40% of the maximum standard uptake value (SUVmax) showed optimal spatial correspondence with the computed tomography‐based gross tumor volume (GTV). The median SUVmax was 6.96, and the Dice similarity coefficient ranged from 0.707 to 0.723. CXCR4‐avid sub‐volumes within GTV were characterized against the sub‐targets of stereotactic body radiotherapy based partial tumor irradiation targeting hypoxic segments (SBRT‐PATHY) and stereotactic centralized ablative radiation therapy (SCART). Dosimetric comparisons were performed between the BG‐pSFRT and SCART plans (15–24 Gy per fraction in three fractions).ResultsCXCR4‐defined sub‐volumes with low thresholds in BG‐pSFRT demonstrated spatial concordance with putative hypoxic sub‐volumes in SBRT‐PATHY, whereas higher SUVmax thresholds yielded multifocal distributions that exhibited spatial discordance relative to SCART, and these multifocal imaging‐defined sub‐volumes were targeted in BG‐pSFRT for dose escalation. Compared to proton SCART, BG‐pSFRT achieved greater high‐dose (V80%), GTV coverage but slightly lower sub‐volume conformity (89.04% vs. 95.55%; P < 0.001), and higher V5Gy and dose per fraction (P < 0.001). All organ‐at‐risk doses met the standard 3‐fraction SBRT constraints.ConclusionsCXCR4‐targeted BG‐pSFRT may be a dosimetrically feasible strategy for bulky limited‐stage small cell lung cancer and demonstrates distinct dosimetric characteristics. The current research findings are limited to the dosimetric level and further clinical trials are required for verification. INTRODUCTION Lung cancer is the most commonly diagnosed cancer and the leading cause of cancer‐related deaths worldwide. 1 Small‐cell lung cancer (SCLC), a highly aggressive subtype, often presents as a bulky disease at diagnosis. 2 , 3 The rapid development of therapeutic resistance limits retreatment efficacy after prior chemoradiation, 4 resulting in a poor prognosis. 3 , 4 , 5 , 6 In limited‐stage disease, a larger tumor volume correlates with worse survival owing to more aggressive biology, increased complications from local therapy, and a microenvironment that promotes heterogeneity and hypoxia. 6 , 7 , 8 , 9 , 10 Stereotactic body radiotherapy (SBRT) is generally not considered appropriate for large‐volume tumors because of limitations in dose conformity and an increased risk of normal tissue toxicity. Hypoxic regions reduce the effectiveness of radiotherapy because hypoxic cells are radioresistant, increasing the risk of treatment failure. 11 , 12 , 13 , 14 Elevated C–X–C motif chemokine receptor 4 (CXCR4) expression is correlated with poor prognosis and treatment resistance in several malignancies. 15 , 16 , 17 Based on this association, the researchers hypothesize that intra‐tumoral regions exhibiting high uptake of the CXCR4‐targeted tracer 1 8F‐AlF‐NOTA‐QHY‐04 may represent niches of relative radioresistance. Thus, targeting CXCR4‐positive sub‐volumes identified by 1 8F‐AlF‐NOTA‐QHY‐04 positron emission tomography (PET) 18 as an attempt to guide spatially fractionated radiotherapy (SFRT) may offer a strategy to improve cytoreduction and shorten treatment duration—an important advantage in small cell lung cancer, where accelerated repopulation occurs during prolonged therapy. 19 This approach aligns with the SBRT‐based partial tumor irradiation targeting hypoxic segments (SBRT‐PATHY), which uses metabolic and perfusion imaging to define the central ablation and peripheral bystander zones. 20 , 21 However, whether CXCR4‐defined resistant regions spatially overlap with the central necrotic/hypoxic areas targeted by SBRT‐PATHY or stereotactic centralized ablative radiation therapy (SCART) remains unclear. In addition, the dosimetric impact of applying SCART or SBRT‐PATHY dose prescriptions to CXCR4‐guided targets via pencil beam scanning proton therapy should be evaluated for safety. This study used a dual‐method approach to address two key questions. First, we characterized the spatial distribution and imaging phenotypes of CXCR4‐avid primary tumors and their subvolumes. Second, we performed a comprehensive dosimetric analysis of proton‐based SFRT plans guided by PET‐derived biological targets. The integrated results provided dosimetric evidence for target specificity and safety, supporting future clinical translation. We term this approach “biologically guided proton SFRT” (BG‐pSFRT). METHODS AND MATERIALS Patient selection We retrospectively analyzed 30 consecutive patients with treatment‐naïve limited‐stage small cell lung cancer who underwent 1 8F‐AlF‐NOTA‐QHY‐04 PET at our institution between November 2022 and July 2024. All procedures adhered to the institutional review board guidelines, and the requirement for informed consent was waived because this retrospective planning study used fully anonymized CT and PET datasets. Staging was performed according the American Joint Committee on Cancer 8th edition. The inclusion criteria were as follows: (1) histologically confirmed limited‐stage small cell lung cancer; (2) age ≥ 18 years; (3) Karnofsky Performance Status ≥ 70; (4) completion of both 1 8F‐AlF‐NOTA‐QHY‐04 PET and contrast‐enhanced CT within two weeks before treatment; and (5) primary gross tumor volume (GTV) on CT > 65 cm3 or short‐axis diameter > 5 cm. The exclusion criteria included prior oncological therapy, distant metastasis, or multiple primary cancers. Patient selection is summarized in the Supplementary Material Figure 1. Spatial correspondence of intra‐tumoral biological sub‐target volumes defined by CXCR4‐targeted PET with SBRT‐PATHY and SCART protocols (representative cases). Panels (a–b) show GTVs (yellow) and hypoxic subvolumes (red) contoured for SBRT‐PATHY; panel (c) shows the SBRT‐PATHY dose distribution; panels (d–e) show GTV (yellow) and BG‐pSFRT (blue) contoured for SCART; panel (f) shows the concept of SCART. CXCR4, C–X–C motif chemokine receptor 4; PET, positron emission tomography; SBRT‐PATHY, stereotactic body radiotherapy‐based partial tumor irradiation targeting hypoxic segments; SCART, stereotactic centralized ablative radiation therapy; GTV, gross target volume; TTV, transitional treatment volume; STV, SCART‐treatment volume. Imaging acquisition protocol The CXCR4‐targeted tracer, 1 8F‐AlF‐NOTA‐QHY‐04, was synthesized using a published method. 18 The patients received an intravenous injection of 4.81 MBq/kg. Whole‐body PET from the head to the thigh was performed approximately 60 min after the injection using a Discovery LS scanner (GE Healthcare, USA). The full‐ring PET scan covered the same axial range without repositioning. Additional CT scans were acquired using uCT906+ (United Imaging, China) and a Biograph Horizon (Siemens Healthineers, Germany) with the following parameters: 120–130 kV, 200–300 mAs, 1‐mm slice thickness. Region of interest delineation Fused PET and CT images were reconstructed in three dimensions and imported into the MIM Software (Cleveland, OH, USA). Two radiation oncologists with > 3 years of experience independently contoured the regions of interest (ROIs), including the primary tumors, on both modalities. On CT, manually contoured GTV included morphologically identifiable tumors and radiologically apparent invasion. No marginal expansion was performed. For PET imaging, PET‐derived GTV were defined using maximum standard uptake value (SUVmax) thresholds of 30%, 35%, 40%, 45%, and 50%. Physiological uptake sites (the liver, kidneys, bladder, and urinary tract) were manually excluded. To determine the optimal SUV% threshold, the spatial overlap between the PET‐derived GTV and manually contoured GTV on CT imaging was assessed using the Dice similarity coefficient. The treatment volume for treatment planning was generated by fusing the manually contoured GTV on the CT images with the PET‐derived GTV, defined by the optimal SUV% threshold using Boolean operations. In SBRT‐based partial tumor irradiation targeting hypoxic segments (SBRT‐PATHY), the hypoxic subvolume encompasses approximately 30% of the GTV through geometric retraction, 21 and; in SCART, a central core of approximately 4.5% of the GTV is defined by isotropic contraction to form a spherical stereotactic centralized target volume. 22 , 23 In proton spatially fractionated radiotherapy (BG‐pSFRT), CXCR4‐defined subvolumes at different levels are generated using SUV thresholds to yield volumes of approximately 30% and 4.5% of the GTV. The CXCR4‐defined subvolumes at different levels on PET imaging were compared with the hypoxic target in SBRT‐PATHY and the central core in SCART. Organs at risk (the lungs, heart, esophagus, and spinal cord) were automatically segmented in RayStation (v12.0.100.0; RaySearch, Sweden) using deep learning and subsequently reviewed manually. The GTV was excluded from the ipsilateral lung parenchyma to enable an accurate dose assessment. Treatment planning Plans were created in RayStation for pencil‐beam scanning using the ProBeam system (Varian, USA). The Ion Monte Carlo algorithm with 1% uncertainty was used for the dose calculation. Biological effectiveness was assumed to be a constant relative biological effectiveness of 1.1 per the International Commission on Radiation Units and Measurements Report 78. 24 A comprehensive robustness evaluation was performed accounting for ± 3.5% range uncertainty and ± 5 mm setup errors. 25 The number of beams (2–4) was individualized based on tumor location and size. The gantry angles were identical between the BG‐pSFRT and SCART plans to isolate dosimetric differences owing to the target definition. Identical optimization objectives and constraints were applied for a fair comparison. Dosimetric evaluation and plan comparison The SCART regimen (15–24 Gy/fraction × 3 fractions) served as a reference for the dosimetric analysis. The comparisons between BG‐pSFRT and SCART addressed the following three domains: First, for target coverage criteria, both plans required D80% to cover ≥90% of central core (in SCART) or CXCR4‐defined subvolumes (in BG‐pSFRT), and 20% prescribed dose to cover ≥ 90% of the treatment volume. The gradient index, 80% prescribed dose volume vs. 20% prescribed dose volume (V80%/V20%) was maintained at > 4.5 to ensure sharp dose fall‐off. Second, for extra‐target dose distribution, dose spillage was evaluated using non‐target volume receiving ≥5 Gy/fraction (V5Gy), and high‐dose spillage near the treatment volume perimeter. Both parameters were minimized during the optimization. Lastly, for dose limits for organs at risk (OARs)., all OAR doses complied with the established clinical guidelines for 3‐fraction SBRT, ensuring both safety and clinical applicability. 26 Prescribed doses were adjusted accordingly. Statistical analysis All analyses were performed using SPSS Statistics (v22.0; IBM, USA). Wilcoxon matched‐pairs signed‐rank tests were used to compare tumor volumes, target coverage metrics, and OAR dose‐volume metrics between the techniques. One‐way repeated measures analysis of variance (ANOVA) was used to assess Dice similarity coefficients, and Spearman's rank correlation analysis was conducted to evaluate the correlations between Dice similarity coefficients and tumor diameter and volume delineated on CT. Statistical significance was set at P < 0.05. RESULTS Of the 48 screened patients with limited‐stage SCLC (Supplementary Material Table 1), 30 with bulky tumors met the inclusion criteria. The cohort included one patient with stage IIA, three with stage IIIA, 14 with stage IIIB, and 12 with stage IIIC. TABLE 1Dice similarity coefficients and correlations with volume and diameter for GTVs from different segmentation methodsManually contoured GTV in CT imagingPET‐derived GTV by SUVmax thresholds (%)3035404550Dice similarity coefficient0.6990.7230.7070.6510.633Correlations with volume−0.067−0.188−0.073−0.0840.063Correlations with diameter0.135−0.0530.039−0.0060.081GTV, gross target volume; CT, computed tomography; PET, positron emission tomography; SUVmax, standardized uptake value (maximum)John Wiley & Sons, Ltd. Dice similarity coefficients and correlations with volume and diameter for GTVs from different segmentation methods GTV, gross target volume; CT, computed tomography; PET, positron emission tomography; SUVmax, standardized uptake value (maximum) On 1 8F‐AlF‐NOTA‐QHY‐04 PET imaging, the median primary tumor SUVmax was 6.96 (range: 5.56–8.92). The PET‐derived GTV using 30%–50% SUVmax thresholds (in 5% increments) was compared to the manually contoured GTV in CT imaging. Volumes at the 35% SUVmax threshold were generally larger than the manually contoured GTV on CT imaging, whereas volumes at the 40% SUVmax threshold were smaller. Wilcoxon signed‐rank tests revealed no significant differences between the groups (P = 0.083 and P = 0.05, respectively). A one‐way repeated‐measures ANOVA revealed significant differences in the Dice similarity coefficients across thresholds. No significant correlations were identified between the Dice similarity coefficient values and the size or diameter of the manually contoured GTV on CT imaging. Consequently, 35%–40% SUVmax thresholds may provide reasonable, although not perfect, spatial overlap with manually contoured GTV in CT imaging despite volume discrepancies. Moreover, the segmentation performance in this range was independent of the tumor size, suggesting its particular utility for bulky small‐cell lung cancer, a common clinical presentation. Figure 1 shows the spatial relationship between the CXCR4‐defined and hypoxic subvolumes in SBRT‐PATHY and the central core in SCART. The CXCR4‐defined subvolumes based on different SUVmax thresholds showed distinct spatial patterns. At approximately 30% of the GTV, these low threshold level subvolumes were located mainly in the central tumor region (Fig. 1a, b), consistent with the hypoxic subvolume in SBRT‐PATHY (Fig. 1c). 21 By contrast, higher SUVmax thresholds yielded multifocal distributions characterized by heterogeneous discrete foci throughout the tumor, designated as CXCR4‐defined subvolumes (Fig. 1d, e). The detailed results are presented in Supplementary Material, Table 2. TABLE 2Dosimetric comparison of the target coverage, dose fall‐off, and OAR doses between BG‐pSFRT and SCART in patients; values are presented as median (IQR); P‐values were calculated using the Wilcoxon matched‐pairs signed‐rank testsBG‐pSFRTSCART P‐valueTargetVolume of 80% prescribed dose (cm3)23.67 (13.44–47.60)21.99 (11.83–34.00)0.00180%/20% (%)11.78 (10.66–13.94)10.40 (9.52–11.38)0.00580% cover of CXCR4‐defined subvolumes / the central core in SCART (%)89.04 (86.05–92.83)95.55 (92.57–96.89)0.00180% cover of GTV (%)11.00 (9.38–13.48)9.30 (8.27–10.29)0.00220% cover of GTV (%)95.04 (88.54–98.98)89.71 (83.88–98.46)0.001Tumor (GTV) edgeV5Gy (cm3)1295.60 (992.23–1715.22)1112.94 (849.27–1450.12)0.001Dose per fraction (Gy)13.94 (12.29–16.01)11.05 (9.97–12.47)0.001Total lung (right plus left)Mean lung dose (Gy)1.85 (1.69–3.19)1.77 (1.57–2.59)0.005V20Gy (cm3)0.06 (0.04–3.18)0.05 (0.03–1.12)0.005Vol. to spare V12.4 (cm3)3326.04 (2688.21–4696.00)3357.34 (2713.46–4733.08)0.001Vol. to spare V11.6 (cm3)3307.13 (2659.73–4663.08)3341.90 (2690.58–4705.13)0.001HeartMaximal dose (Gy)27.47 (19.72–28.60)17.74 (15.17–21.94)0.001V24Gy (cm3)0.08 (0–0.19)0 (0–0.01)0.008OAR, organ at risk; IQR, interquartile range; BG‐pSFRT, biologically guided proton spatially fractionated radiotherapy; SCART, stereotactic centralized ablative radiation therapy; GTV, gross target volumeJohn Wiley & Sons, Ltd. Dosimetric comparison of the target coverage, dose fall‐off, and OAR doses between BG‐pSFRT and SCART in patients; values are presented as median (IQR); P‐values were calculated using the Wilcoxon matched‐pairs signed‐rank tests OAR, organ at risk; IQR, interquartile range; BG‐pSFRT, biologically guided proton spatially fractionated radiotherapy; SCART, stereotactic centralized ablative radiation therapy; GTV, gross target volume Figure 2 summarizes the target volume parameters used in this study, with SUVmax thresholds (60%–84%, mean: 76%) indicating the threshold applied to generate CXCR4‐defined subvolumes. The dosimetric comparison focused on cases with substantial spatial discordance but comparable CXCR4‐defined subvolumes and spherical stereotactic centralized target volumes in SCART, enabling the direct evaluation of divergent targeting strategies. Geometric parameters of GTV and CXCR4‐defined subvolumes, including diameter, GTV, CXCR4‐defined subvolumes, CXCR4 subvolumes to GTV ratio, and the SUVmax‐based threshold (expressed as %SUVmax). GTV, gross target volume; SUVmax, standardized uptake value (maximum); CXCR4, C–X–C motif chemokine receptor 4. The dosimetric parameters for the target coverage, dose fall‐off, and OAR doses are summarized in Table 2. BG‐pSFRT achieved significantly higher V80% and V80%/V20% ratio than SCART (P < 0.001), indicating broader high‐dose ablation within the GTV. It also provided superior GTV coverage at both 80% and 20% prescription doses. However, CXCR4‐defined subvolume coverage by an 80% dose was lower in BG‐pSFRT than in the central core coverage in SCART (89.04% vs. 95.55%; P < 0.001), reflecting a trade‐off between creating a heterogeneous lattice and conforming to small, multifocal targets. Additionally, BG‐pSFRT resulted in higher V5Gy and maximum dose to peritumoral normal tissues (P < 0.001). Although BG‐pSFRT showed slightly higher OAR doses, all plans met the standard 3‐fraction SBRT constraints, demonstrating successful optimization balancing efficacy and safety. Both techniques showed elevated peripheral doses owing to high prescription levels. The representative dose distributions are shown in Figure 3. In BG‐pSFRT (a), the high‐dose region conformed to CXCR4‐defined radioresistant subvolumes, forming a “lattice”‐like pattern. In SCART (b), the dose distribution resembled that of conventional SBRT, with a concentric radially decreasing gradient from the central zone to the periphery. Representative multi‐planar dose distributions for (a) BG‐pSFRT and (b) SCART in one patient. Axial and sagittal/coronal views are shown; the color wash indicates isodose levels (percentage of the prescription dose) as shown in the scale bar. BG‐pSFRT, biologically guided proton spatially fractionated radiotherapy; SCART, stereotactic centralized ablative radiation therapy. DISCUSSION The management of limited‐stage SCLC has long relied on thoracic radiotherapy at 45 Gy in 30 twice‐daily fractions 27 , 28 and dose escalation to 54–60 Gy improves overall survival. 29 , 30 , 31 SBRT use is often limited by rapid tumor proliferation and frequent presentation of bulky masses at diagnosis. 19 , 32 SFRT techniques have shown potential for rapid cytoreduction and improved local control in preclinical and clinical studies by leveraging dose‐volume effects, immunomodulation, and intratumoral heterogeneity. 33 , 34 , 35 Building on the evidence that the CXCR4 marks treatment‐resistant regions in SCLC, 36 , 37 a novel BG‐pSFRT is proposed in this study. It leverages 1 8F‐AlF‐NOTA‐QHY‐04 PET imaging to define functional subvolumes within primary SCLC tumors, alongside pencil‐beam scanning proton therapy for precise dose modulation. 23 It is critical to understand that CXCR4‐PET was utilized here as a hypothesis‐generating tool for outlining biologically distinct intratumoral regions for dose painting, and not as a validated surrogate marker for radioresistance, as the biological interpretation of CXCR4‐avid subvolumes remains speculative. 1 8F‐AlF‐NOTA‐QHY‐04 effectively visualizes GTV size and location on PET, with an optimal SUVmax threshold of 35%–40%, similar to the 40% reported in non‐small‐cell lung cancer. 38 PET imaging revealed two distinct subvolumes within the GTV: a smaller region (approximately 4.5% of the GTV) defined by high SUVmax thresholds showed a multifocal, three‐dimensional distribution (Figs. 1d–e), unlike the central target at the GTV center in SCART (Fig. 1f). 22 , 23 This nonuniform spatial configuration underscores the potential of molecular imaging to reveal tumor heterogeneity that is invisible to anatomy alone. In contrast, biologically guided approaches have been developed. Unlike SCART's reliance solely on geometric parameters, SBRT‐PATHY's targets inferred from hypoxia based on conventional 1 8F‐FDG tracers (which lack specificity for intra‐tumoral hypoxia), 20 , 21 , 39 or MRI‐based BG‐LRT's geometric lattice adjustments, 40 we use a putative molecular target (CXCR4) for dose‐painting. However, the clinical significance of targeting specific multifocal CXCR4‐high regions requires further investigation. Furthermore, although 1 8F‐FDG primarily reflects glucose metabolism, which is often globally elevated in rapidly growing tumors, CXCR4 expression may highlight specific populations of tumor cells or microenvironments associated with invasiveness, metastasis, or radioresistance. 15 , 16 , 17 , 36 , 37 It is plausible that CXCR4‐avid regions overlap with areas of high metabolic activity identified by FDG‐PET, particularly in aggressive tumor foci. However, given CXCR4's role in pathways distinct from glucose uptake, it could also delineate subvolumes that are not intensely FDG‐avid, but are critical for treatment resistance or disease progression. A direct comparison in future studies is essential to elucidate these potential commonalities and distinctions, thereby clarifying the unique contribution of CXCR4‐targeted imaging to biologically guided radiotherapy. Both SCART and BG‐pSFRT generate distinct high‐, intermediate‐, and low‐dose regions within the GTV; however, their spatial configurations differ significantly. The BG‐pSFRT produced a heterogeneous lattice‐like pattern with high‐dose peaks in the axial plane (Fig. 3a). Compared to SCART, it resulted in higher V5Gy per fraction and slightly increased peripheral GTV dose due to higher prescription doses. The immunomodulatory effect of SFRT depends on a balance: high “peak” doses induce immunogenic cell death, although lower “valley” doses preserve antigen‐presenting and immunocompetent cells in the peritumoral stroma, 33 , 34 warranting careful optimization to maintain this therapeutic window. All risk doses met the standard 3‐fraction SBRT constraints. However, oncologists and medical physicists need to clearly recognize the significant disparity in tumor size between SBRT and SFRT, and the resulting established dose constraints for at‐risk organs may not be suitable. Proton therapy improves the organs at risk, enabling full spinal cord avoidance and reducing the esophageal dose. The advantages of BG‐pSFRT presented here are primarily technical and dosimetric: (1) the use of a novel targeted tracer allowing visualization of a specific molecular phenotype, (2) the ability to precisely deliver ablative doses to these imaging‐defined regions, (3) greater high‐dose volume coverage within tumors compared to traditional methods, and (4) a promising investigation strategy for bulky limited‐stage SCLC beyond conventional chemo‐radiotherapy. Unlike SCART or SBRT‐PATHY, BG‐pSFRT provides a novel platform for testing the biological principles of targeting molecularly defined niches in bulky tumors. Combined with proton pencil beam scanning, it enhances dose modulation and targeting accuracy in high‐dose regions. 23 When applied to limited‐stage SCLC using CXCR4‐targeted tracers, BG‐pSFRT shows promising clinical potential for further investigation and validation. For such complex plans, verification must consider target biology, dosimetry, and proton beam physics. 41 , 42 The study limitations include the (1) lack of biological validation for the CXCR4‐avid subvolumes (e.g., immunohistochemistry for CXCR4, hypoxia markers or stem cell markers); (2) lack of a direct comparison between 1 8F‐AlF‐NOTA‐QHY‐04 and 1 8F‐FDG; (3) the fluorine‐18‐labeled tracer is restricted to imaging and diagnostic applications, in contrast to isotopes such as gallium‐68 and lutetium‐177 that enable concurrent theranostic use; (4) a narrow focus on a single disease—limited‐stage SCLC—and a single modality (PET imaging); and (5) the limited sample size and absence of large‐scale, multicenter validation data constrain the generalizability of our findings. Future studies should incorporate multimodal imaging approaches, such as PET/MRI with diffusion‐weighted imaging, extend their application across diverse cancer types, and further confirm their utility with multicenter validation data to enhance biological target definition and its generalizability. This favorable dosimetric safety profile provides a promising rationale for initiating phase I trials to evaluate the feasibility and therapeutic potential of BG‐pSFRT. CONCLUSION BG‐pSFRT—integrating CXCR4‐targeted tracer 1 8F‐AlF‐NOTA‐QHY‐04 PET imaging with proton‐based spatially fractionated radiotherapy—may offer a feasible and innovative approach for treating bulky limited‐stage small cell lung cancer. Its dosimetry supports its preclinical safety and warrants further evaluation in clinical trials. Future efforts should focus on refining biological target definitions, elucidating the dose–response relationship for immune activation, and developing spatially fractionated radiotherapy‐specific normal tissue dose limits. AUTHOR CONTRIBUTIONS Keru Pu: Writing — original draft, conceptualization, and validation. Tengxiang Li: Writing — original draft, conceptualization, methodology, and validation. Shengnan Xu: Investigation and validation. Xue Sha: Methodology. Qingtao Qiu: Methodology and validation. Jinming Yu: Writing — review and editing, supervision, and methodology. Yong Yin: Writing — review and editing, supervision, and project administration. Jie Liu: Writing — review and editing, project administration, and funding acquisition. CONFLICT OF INTEREST STATEMENT All authors declare no potential conflicts of interest. ETHICAL STATEMENT This study exclusively used de‐identified retrospective datasets. According to institutional policies and prevailing national regulations, retrospective analyses involving anonymized data are exempt from formal ethical committee approval.