Volumetric and spatial changes in brain metastases during stereotactic radiosurgery: a case for off-line adaptive planning

Highlights

• •Tumor displacement during FSRT may compromise target accuracy.
• •Interim MRI can guide offline adaptive planning in selected brain SRS cases.
• •Adaptive SRS should be considered in patients with edema and midline shift.

Abstract

Stereotactic radiosurgery is a widely used treatment modality for brain metastases, particularly in cases with a limited number and volume of lesions. While adaptive strategies have gained prominence in the stereotactic treatment of extracranial, their application in intracranial SRS remains largely unexplored. This case study presents a 45-year-old female with multiple brain metastases and significant perilesional edema, treated with fractionated stereotactic radiotherapy. Due to tumor displacement during the treatment course, adaptive SRS was implemented after the third fraction, utilizing an updated MRI scan to account for changes in tumor volume and midline shift. The total GTV decreased by 34.2 %, and five lesions were found to be partially outside the prescribed treatment target. Based on this MRI, an offline adaptive SRS plan was generated, and the remaining two fractions were delivered according to the updated plan. This case underscores the potential for significant changes in target volume and spatial displacement during FSRT in patients with brain metastases exhibiting extensive edema and emphasizes the importance of mid-treatment imaging and the potential role of offline adaptive SRS strategies in managing intracranial tumors.

Introduction

Stereotactic radiosurgery (SRS) is an effective and widely used treatment modality for brain metastases, particularly in cases with a limited number and volume of lesions [1,2]. Advances in radiotherapy, including image-guided radiation therapy (IGRT), motion management, stereotactic immobilization, high-resolution imaging of target volumes, on-board imaging for intra-fraction and inter-fraction adjustments, precise high-dose delivery, and adaptive strategies, have significantly improved the precision and feasibility of stereotactic treatments [[3], [4], [5]].

Adaptive radiotherapy (ART) can be classified as either online or offline [6]. Offline adaptive radiotherapy is predominantly employed in long-course treatments, particularly in lung and head and neck cancers, where tumor shrinkage, weight loss, or anatomical alterations necessitate treatment modification [[7], [8], [9]]. While adaptive strategies have gained prominence in the stereotactic treatment of extracranial malignancies especially in online setting, their application in intracranial SRS remains largely unexplored [10,11]. The limited data on adaptive SRS for brain metastases may be attributed to several factors. Traditionally, brain SRS has been performed as a single-session treatment using Gamma Knife. With the introduction of linear accelerators, fractionated stereotactic radiotherapy (FSRT) has enabled treatment completion within 3–5 fractions over a short timeframe. Moreover, the likelihood of significant anatomical changes during treatment remains relatively low. Additionally, the availability of IGRT techniques and the challenges in detecting the need for adaptation in small lesions using suboptimal imaging modalities—such as 2D kV imaging in CyberKnife or cone beam computed tomography (CBCT) on conventional C-arm linear accelerators, which cannot adequately differentiate soft tissue changes—further limit the application of adaptive strategies.

Although dynamic changes in resection cavity volumes following metastasectomy are well recognized, data regarding adaptive FSRT in the context of intact brain metastases remain scarce [[12], [13], [14]]. The necessity for adaptive SRS in such cases may be infrequent; however, it may be crucial in specific clinical scenarios.

This case highlights the potential need for ART planning in patients with significant perilesional edema and midline shift who are receiving anti-edema therapy. In particular, for patients with multiple brain metastases, tumor displacement during the treatment course may result in suboptimal targeting and excessive radiation exposure to healthy tissues. This could, in turn, impact local control rates and increase radiation-induced toxicity.

Case

A 45-year-old female patient was initially diagnosed with triple-negative invasive breast carcinoma in September 2020 after presenting with a palpable mass in the right breast. A biopsy confirmed the diagnosis, and an axillary lymph node biopsy with axillary marking was performed. The patient underwent neoadjuvant chemotherapy with FEC (fluorouracil, epirubicin, cyclophosphamide) for four cycles followed by docetaxel for four cycles. On May 18, 2021, the patient underwent a right breast conserving surgery with axillary dissection. Pathological evaluation revealed ypT0N2 disease, with 6 out of 19 lymph nodes involved (five with macrometastases >2 mm, one with micrometastases < 2 mm). A positron emission tomography–computed tomography (PET-CT) scan in May 2021 detected residual level III axillary lymph node involvement. Postoperative radiotherapy was administered between June 7, 2021, and July 16, 2021. Adjuvant chemotherapy with carboplatin and capecitabine was completed in October 2021.

In June 2022, disease progression was observed with mediastinal lymph node involvement, leading to the initiation of pembrolizumab, which was continued thereafter.

In March 2023, the patient presented to the emergency department with a one-month history of progressive cognitive impairment, headaches, nausea, and vomiting. Brain magnetic resonance imaging (MRI) revealed 10 brain metastases with perilesional edema causing a minimal midline shift. Dexamethasone (16 mg/day) was initiated, and FSRT was recommended. To account for potential plan adaptation and midline shift regression due to anti-edema treatment, pre-treatment verification imaging with control brain CT scans was performed before each SRS fraction. Treatments were planned based on T1 contrast enhanced (magnetic resonance imaging) MRI. A plan was made to deliver a total dose of 30 Gy over 5 days, with 6 Gy administered daily. The patient was treated with robotic CyberKnife® (Accuray Incorporated, Sunnyvale, CA based on our previously published institutional brain metastases treatment protocol [15]. By the third fraction, a reduction in midline shift and tumor displacement was noted, necessitating an adaptive replanning approach. A new contrast-enhanced brain MRI demonstrated regression in lesion size and edema, as well as midline shift regression. Based on this MRI, an offline adaptive SRS plan was generated, and the remaining two fractions were delivered according to the updated plan (Fig. 1). The patient was treated with the adaptive plan for the last two fractions. The total gross tumor volume (GTV) has decreased by 34,2 %. Furthermore, it was observed that five of the lesions were partially outside the prescribed treatment target. The comparison of the volumetric and dosimetric parameters of the treatment plans were summarised in Table 1. The treatment was administered from March 15 to March 23, 2023. Dexamethasone was tapered and discontinued after treatment.

Fig. 1.

Volumetric and Spatial Changes in tumor GTVs. The GTV volumes delineated on the pre-planning MRI are shown on contrast-enhanced T1 VIBE brain MRI images acquired after the third fraction. Displacement of small GTVs, indicated by the blue arrow, is observed due to the reduction in brain shift, resulting in a partial target miss. Tumor volume reduction is indicated by the red arrows. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1.

Volumetric and Dosimetric comparison of the treatment plans.

Parameter	Baseline Plan	Adaptive Plan
Total GTV volume (cc)	20.54	13.52
GTV1	0.03	0.03
GTV2	0.04	0.03
GTV2	0.07	0.04
GTV3	0.09	0.06
GTV4	0.16	0.1
GTV5	0.56	0.33
GTV6	0.72	0.46
GTV7	2.21	1.17
GTV8	3.69	2.39
GTV9	4.98	3.24
GTV10	7.99	5.67
Brain-GTV doses
7 cc (Gy)	28.8	28.4
16 cc (Gy)	26.5	25.7
V24Gy (cm3)	27.91	22.71
Right Hippocampus (mean)(Gy)	5.63	5.89
Left Hippocampus (mean) (Gy)	11.35	8.23
Chiasm (max) (Gy)	4.8	7.2
R Optic nerve (max) (Gy)	4	7.2
L Optic Nerve (max) (Gy)	4.83	7.2
Brainstem (max) (Gy)	9.78	15.24

Following SRS, sacituzumab govitecan-hziy was initiated. After SRS, patients underwent follow-up with clinical and radiographic surveillance with brain MRIs at 3 months after SRS and subsequently every 3 months thereafter. First follow-up brain MRI in June 2023 demonstrated local control in all treated lesions with no new metastases. However, in October 2023 and February 2024, the patient underwent additional SRS sessions due to newly emerging brain metastases. In April 2024, leptomeningeal disease was diagnosed. The patient was lost due to the disease on June 24, 2024.

Discussion

In this case, among the 10 brain metastases in a patient treated with FSRT, a reduction in tumor volume was observed in 9 of them (excluding the smallest lesion) following three fractions of treatment, as evidenced by brain MRI. The total GTV volume demonstrated a decrease of 34.2 %. Due to the regression of the shift resulting from the reduction in brain edema, tumor displacement was observed, leading to five lesions being partially located outside the prescribed treatment target. This case highlights the potential for significant changes in target volume and spatial displacement during FSRT in patients with brain metastases exhibiting extensive edema and midline shift. We emphasize the necessity of volumetric imaging (preferably MRI) prior to treatment to assess the need for ART in such cases.

Contrast-enhanced brain MRI is routinely utilized in SRS planning for brain metastases [16]. The impact of imaging timing on SRS treatment outcomes has been previously established, with documented changes in tumor volume, number, and spatial displacement. Seymour et al. reported that delaying treatment beyond 14 days from the planning MRI was associated with a lower local progression-free survival (hazard ratio [HR]: 3.4, 95 % confidence interval [CI]: 1.6–7.3 0) [17]. Furthermore, Salkeld et al. conducted a prospective analysis comparing standard brain MRI with a treatment planning MRI performed within 2 4 h before treatment [18]. Their findings demonstrated that more than 50 % of patients required a change in management, including modifications in target volume, lesion count, or spatial displacement. When the interval exceeded seven days, 78 % of patients experienced changes in management, compared to 41 % for those scanned within a seven-day period. Similarly, Jarvis et al. analyzed 41 patients (43 surgical cavities) and found significant volumetric changes (>2 cc) in 53.5 % of cases, with 23.3 % showing cavity collapse and 30.2 % demonstrating increased volume. Notably, spatial displacement was observed in 66.8 % of cases, with 12.8 % experiencing shifts greater than 1 mm [19]. Additionally, Hessen et al. reported a tumor displacement of 1.3 mm related to edema changes [20].

For intact brain metastases, Uto et al. analyzed 27 lesions in 21 patients receiving hypofractionated treatment (39 to 44.2 Gy in 13 fractions) with a median tumor volume of 3.8 cc [21]. Mid-treatment MRI was performed for target lesion evaluation, revealing a >20 % increase in five lesions and a >20 % decrease in another five. Notably, more than half of the lesions had an adapted GTV that did not overlap with the original PTV. In our case, re-evaluation and re-contouring during the fourth fraction revealed a significant decrease in GTVs. In a separate case study, Rosa et al. reported a case of a patient with isolated brain metastasis treated with MR-Linac-guided fractionated SRS in three fractions. A T1 Gadolinium-enhanced sequence obtained prior to the second fraction revealed a GTV increase of 52.3 %, with a DICE similarity index of only 0.77 [22]. Based on the evaluation of anatomical and dosimetric deviations from the initial plan, offline plan adaptation was performed to improve target coverage for the third and final fraction. These studies highlight that during fractionated SRS for intact brain metastases, treatment volumes may change—either increasing or decreasing—based on mid-treatment MRI findings, and adaptive planning may be required accordingly. The relevant studies are summarized in detail in the Table 2.

Table 2.

Summary of studies evaluating mid-treatment volumetric changes during fractionated SRS for intact brain metastases.

Study (Year)	Patients / Lesions	Primary Histology	Fractions	MRI Evaluation (Fraction)	Imaging Modality	GTV Volume Change	Adaptation Performed
Uto et al. [21] (2021)	23 / 27	Mixed (lung, breast, GI)	13	6th (range: 3rd–11th)	Diagnostic MRI (field strength NR)	>20 % increase in 5 lesions; >20 % decrease in 5 lesions	Not reported
Rosa et al. [22] (2024)	1 / 1	NSCLC	3	Prior to 2nd fraction	MRIdian – T1 Gadolinium-enhanced	52.3 % increase	Offline adaptation for 3rd fraction
Current study	1 / 1 0	Breast	5	Prior to 4th fraction	Diagnostic MRI (3T)	34.2 % decrease	Offline adaptation in 4th fraction

With the integration of MR-Linac technology into clinical practice, daily MRI acquisitions during treatment have enabled the assessment of tumor dynamics over time. The possibility of daily MRI-based adaptive SRS has been explored, particularly in patients with large brain metastases or post-resection cavities. For resection cavities, Tan et al. evaluated inter-fractional dynamics in 15 patients undergoing a five-fraction MR-Linac-based treatment (1.5T MR-Linac; Unity, Elekta, Stockholm, Sweden) with a total dose of 27.5 to 30 Gy [23]. They utilized T1-enhanced sequence imaging on the simulation day and in the third fraction, alongside T2/FLAIR sequence imaging performed on the simulation day and throughout all five fractions. Their findings revealed median relative reductions in target volume of 11.4 % and 8.4 % for the respective sequences in fraction three compared to the simulation day. A study investigating MR-Linac-based adaptive treatment in eight patients with 20 large brain metastases assessed daily MR imaging over five fractions, demonstrating an average tumor volume reduction of 28.8 % (range, 10.0 % to 55.0 %) [24]. Another dosimetric MR-Linac study by Snyder KC reported improvements in target coverage and normal brain sparing with online adaptive MR-guided FSRS [25]. While MR-Linac systems enable online ART for shrinking or shifting lesions, the lack of non-coplanar beam arrangements and six degrees of freedom (6DOF) corrections in the treatment table may impose dosimetric limitations, making online adaptation not always the optimal approach. However, the ability to perform daily MR imaging offers valuable insights into tumor volume changes, particularly in large tumors treated with fractionated SRS, underscoring the need for adaptation—potentially best achieved through offline strategies. These findings highlight the potential need for offline adaptation strategies in fractionated SRS to account for inter-fractional anatomical variations.

Conclusion

This case highlights that tumor volume and location may change during brain SRS, potentially necessitating ART. It is crucial to identify which patients would benefit from adaptive brain SRS, assess its clinical significance, and consider the additional workload associated with its implementation. Large-scale prospective studies are needed to establish objective criteria for selecting patients who require ART.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.