A Study of Pseudoprogression after Spine SBRT

Purpose: To determine the incidence of pseudoprogression (P P) post-spine SBRT based on a detailed and quantitative assessment of MRI morphol ogic tumor alterations, and to identify predictive factors distinguishing PP from local rec urrence (LR). Materials/Methods: A retrospective analysis of 35 patients with 49 sp inal segments treated with spine SBRT, from 2009 to 2014, was conducted. Media n number of follow-up MR studies was 4 (range: 2-7). The gross tumor volumes (GTV) within each of the 49 spinal segments were contoured on the pre-treatment and each subsequent follow-up T1 and T2 weighted MRI sagittal sequence. T2 signal intensity was reported as the m ean intensity of voxels constituting each volume. LR was defined as persistent GTV enlargeme nt on at least 2 serial MRI for at least 6 months, and/or upon pathological confirmation. PP w as defined as a GTV enlargement, followed by stability or regression on subsequent imaging wi thin 6 months. Kaplan Meier analysis was used for estimation of actuarial LC, disease-free s urvival (DFS) and overall survival (OS). Results: Median follow-up was 23 months (1-39 months). PP w as identified in 18% (9/49) of treated segments, and LR in 29% (14/49). Earlier vo lume enlargement (5 months for PP vs. 15 months for LR, p=0.005), greater GTV to reference n on-irradiated vertebral body (VB ref) T2 intensity ratio (+30% vs. -10% for LR, p=0.005) and growth confined to the 80% prescription isodose line (IDL) (8/9 vs. 1/14, p = 0.002) were a ssociated with PP on univariate analysis. Multivariate analysis confirmed an earlier time to v lume enlargement and growth within 80% IDL as significant predictors of PP. LR involved t he epidural space in all but 1 lesion, whereas PP was confined to the VB in 7/9 cases. Conclusions: PP was observed in 18% of treated spinal segments. Tumor growth confined to the 80% IDL and earlier time to tumor enlargement predi cted for PP. M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT INTRODUCTION Spine stereotactic body radiation therapy (SBRT) is an emerging therapy for patients with spinal metastases. The high in-field biologically effectiv e dose (BED) and steep dose gradients allow for delivery of a highly focused ablative dose. Unc o trolled data suggest greater rates of complete response to pain, local control and possib ly neurological function as compared to conventional external beam palliative radiation [17]. This technique is still considered emerging and several questions remain as to the optimal dose and fractionation, tolerance to the critical organs-at-risk (OAR) and, moreover, how to judge tr eatment response. The latter is of significant importance as it is only recently that we have rout inely incorporated MRI into the treatment planning and follow-up of spinal metastases treated with radiation. As suggested by the recently reported SPIne response assessment in Neuro-Oncolog y (SPINO) group [8], MRI is the recommended imaging modality for diagnostic evaluat ion and assessment of tumor response following spine SBRT [9]. This is due to the abilit y o visualize the tumor within the bony segment as opposed to a more crude delineation limi ted to the bony anatomy possible with CT. The impact of post radiation MRI signal change has only begun to be investigated, and as a direct result of SBRT practice that demands rigorous follo w-up given the potential for serious complications like radiation myelopathy [10] and ve rtebral compression fracture (VCF) [11]. Moreover, when applying such a high dose technique, documentation and understanding of treatment response is imperative to reassure the pa ti nt and also for the field to evolve. At present, the morphologic changes associated with ra diation in the spine remains poorly understood [8]. PP is defined as a treatment-related transient tumo r growth that mimics true progression [12,13]. It was first described in gliomas undergoing high d ose radiation and chemotherapy [12,14-16], and has been well documented following brain radios urgery [17-22], lung SBRT [16,23-26] and M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT liver SBRT [27]. However, literature on osseous PP post spine SBRT is limited and although its occurrence was suggested in some early spine SBRT l iterature [28,29], it was only recently that Taylor et al. [30] reported a detailed summary of 2 patients that experienced imaging-based PP. Most recently, the MD Anderson Cancer Center group reported the occurrence of PP in 5 of 37 lesions treated with spine SBRT [31]. They suggeste d that PP should be considered before assuming local relapse (LR) and that serial imaging should be considered; however, their analysis did not focus on quantitative MR based signal chara cte istics. In the spine, the consequence of misdiagnosing LR can be significant, as often the n ext step in the management of SBRT failures is a spine surgery which has risks of morbidity. Fu rthermore, if a second course of radiation is delivered assuming LR, the patient may be at seriou s risk of irreversible devastating toxicities that include radiation myelopathy. A clear underst anding of response is needed in order to make appropriate treatment decisions, in particular give n the potential for PP. The aim of this study was to determine the incidence of PP post-spine SBRT, b ased on a detailed and quantitative assessment of MRI morphologic tumor alterations, as well as to identify predictive factors distinguishing PP from LR. MATERIAL AND METHODS Patient selection From July 2009 to March 2014, 127 patients were tre ated with spine SBRT at our center. Patients were included in this analysis if they had a pre-tr eatment MRI demonstrating spinal metastatic disease and at least 2 post-treatment follow-up MRI studies. Patients undergoing systemic therapy such as cytotoxic chemotherapy, hormone the rapy or bisphosphonates could be included in the study if they were stable on treatment based on the baseline assessment. This left 35 M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT patients with 49 spinal segments as the study cohor t. Institutional ethics review board approval was obtained for this study. In general, our institutional indications for spine SBRT include treatment of oligometastatic disease, re-treatment after conventional radiothera py nd treatment of residual disease after surgical debulking. Patients with spinal cord compr ession or a mechanically unstable spine were only treated with SBRT after surgical decompression or stabilisation. Radiosensitive histologies including lymphoma, seminoma and multiple myeloma w ere excluded. A radioresistant histology was defined as those metastases arising from thyroi d, enal, melanoma or sarcomas. All other histologies were classified as neutral. Epidural di sease grading was based on Bilsky criteria [32], and extra-compartmental vs. intra-compartmental dis ease classification was categorized as per the Tomita guideline [33]. SBRT treatment Patients were treated with either Cyberknife (XSigh t Spine, Cyberknife (G4 2009-2013, VSI 2013-2014), Accuray, Sunnyvale, CA), linac-based vo lumetric intensity-modulated radiotherapy (RapidArc®, Varian Medical Systems, Palo Alto, CA) (VMAT) or helical tomotherapy (HT) (Tomotherapy Hi-Art, Accuray, Sunnyvale, CA). Immob ilization devices included a custom foam cushion for Cyberknife Xsight Spine® based SBR T. A BodyFix (Elekta AB, Stockholm, Sweden) was used for HT and VMAT based SBRT for spi nal segments involving the thoracic 4 th vertebrae (T4) and lower; otherwise, a thermoplasti c head and neck mask was used for spinal segments involving the 3 rd thoracic vertebrae to the 1 st cervical vertebrae. Pre-treatment contours for all treatment modalities were performed on Ecli pse treatment planning system (Varian Medical Systems, Palo Alto, CA). All patients underwent a 1.5 mm slice thickness non -c trast planning CT-scan in the supine M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT position. Sagittal non-enhanced 3D T1 SPACE (TR 420 ms, TE 14 ms, FOV adjusted to volume of interest, slice thickness 1 mm, voxel size 1.0x1 x1.0 mm) and 3D T2 SPACE (TR 1200 TE 154 FOV adjusted to volume of interest, voxel size 1.0x1.0 x1.0 mm). MRI sequences acquired on a 1.5 Tesla system (Aera, Siemens, Forchheim, G ermany) extending cranio-caudally by at least one vertebral body below and above the treate d lesion were co-registered with the planning CT scan. The fused images were used to delineate tu mor and spinal cord. The CTV was based on the discretion of the treating radiation oncologist and typically involved at least a 5 mm margin beyond the GTV until 2012. Thereafter, the Internat ional Spine Radiosurgery Consortium (ISRC) consensus guidelines for target volume definition w ere implemented as standard of practice. A planning target volume (PTV) margin of 2 mm was app lied excluding the spinal cord and/or thecal sac, and a 2mm planning risk volume (PRV) wa s applied to the spinal cord as a safety margin. Dose restrictions to spinal cord were per ublished constrains [10,34]; limitations for other OAR as per RTOG 0631 [35]. Follow-up, image analysis and statistics Standard follow-up for all patients included a medi cal visit and a non-contrast enhanced MRI every 2 to 6 months. MRIs were performed on a 1.5T system (Avanto, Siemens, Forchheim, Germany), different from the treatment planning MRI system. Follow-up studies were on varied 1.5T devices, and routinely included non-enhanced T 1 turbo Spin Echo (TR 685 ms, TE 13 ms, FOV 330 mm, slice thickness 3.5 mm, interslice gap 0.3 mm) and T2 Turbo Spin Echo (TR 4330 ms, TE 83 ms, FOV 330 mm, slice thickness 3.5 mm, i nterslice gap 0.3 mm) MRI sequences. Pre-treatment and follow-up sagittal T1and T2-wei ghted sequences were imported into a single treatment planning system (Eclipse, Varian Medical Systems, Palo Alto, CA). Bony registration M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT between pre-treatment and follow-up MRI was perform ed on