Heterogeneity in treatment response of spine metastases to spine ...

Accepted Manuscript Heterogeneity in treatment response of spine metastases to spine stereotactic radiosurgery within “radiosensitive” subtypes Vincent Bernard, MS, Andrew J. Bishop, MD, Pamela K. Allen, PhD, Behrang Amini, MD, PhD, Xin A. Wang, PhD, Jing Li, MD, PhD, Claudio E. Tatsui, MD, Laurence D. Rhines, MD, Paul D. Brown, MD, Amol J. Ghia, MD PII:

S0360-3016(17)33784-7

DOI:

10.1016/j.ijrobp.2017.08.028

Reference:

ROB 24474

To appear in:

International Journal of Radiation Oncology • Biology • Physics

Received Date: 8 October 2016 Revised Date:

11 August 2017

Accepted Date: 21 August 2017

Please cite this article as: Bernard V, Bishop AJ, Allen PK, Amini B, Wang XA, Li J, Tatsui CE, Rhines LD, Brown PD, Ghia AJ, Heterogeneity in treatment response of spine metastases to spine stereotactic radiosurgery within “radiosensitive” subtypes, International Journal of Radiation Oncology • Biology • Physics (2017), doi: 10.1016/j.ijrobp.2017.08.028. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Heterogeneity in treatment response of spine metastases to spine stereotactic radiosurgery within “radiosensitive” subtypes

RI PT

Vincent Bernard, MS*, Andrew J. Bishop, MD+, Pamela K. Allen, PhD+, Behrang Amini, MD, PhD^^, Xin A. Wang , PhD++, Jing Li, MD, PhD+, Claudio E. Tatsui, MD^, Laurence D. Rhines, MD^, Paul D. Brown, MD**, Amol J. Ghia, MD+

SC

*The University of Texas Health Science Center, Graduate School of Biomedical Sciences, +Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, ++ Department of Radiation Physics, The University of Texas MD Anderson Cancer Center, Houston, Texas, ^Department of Neurosurgery, The University of Texas MD Anderson Cancer Center, Houston, Texas ^^Department of Diagnostic Radiology, The University of Texas MD Anderson Cancer Center, Houston, Texas **Department of Radiation Oncology, Mayo Clinic

M AN U

Disclosures: None

AC C

EP

TE D

Acknowledgements This work was supported by the Cancer Prevention Research Institute of Texas [RP140106 to V.B.].

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Summary Spine stereotactic radiosurgery (SSRS) is an effective treatment of metastatic disease, yet therapeutic response among classically defined radiosensitive histologies is heterogeneous. Following a retrospective review of patients treated within our institution, we identified non-small cell lung (NSCLC) and colorectal cancers as radiosensitive subtypes at a high risk for local failure. This work suggests a personalized approach to SSRS based on histology, supporting the role of dose escalation in NSCLC/colorectal cancer histologies.

ACCEPTED MANUSCRIPT

Heterogeneity in treatment response of spine metastases to spine stereotactic radiosurgery within “radiosensitive” subtypes

RI PT

Abstract Purpose: To characterize outcomes of metastatic colorectal and non-small cell lung cancer (NSCLC) histologies, compared to other radiosensitive histologies, treated with

SC

spine stereotactic radiosurgery (SSRS) with regards to local control, overall survival, and

M AN U

predictors of response.

Methods and Materials: A total of 127 patients with 148 SSRS treated metastatic lesions at our institution between 2003 and 2013 were reviewed. We assessed clinical endpoints in relation to histological type including local control (LC) and overall survival

TE D

(OS) using univariate and multivariate analyses.

Results: For all patients, the 1 and 2 year actuarial rates for LC were 82.6% and 75.8%; and rates for OS were 72.9% and 51.5% respectively. Among tumor histologies, 1 year

EP

cumulative incidence rates (CIR) of local failure for thyroid, breast, lung, and colon cancer were 8.7%, 7.0%, 26.6%, and 39.6% respectively. When analyzed together,

AC C

NSCLC and colorectal cancers had significantly greater CIR at 1 and 2 years (30.4% and 38.7%, respectively), than other histologies (8.0% and 14.1% respectively, p=0.0008). NSCLC/colorectal tumor status was a significant predictor of local failure in a competing risk univariate model (HR 2.12, 95% CI 1.07 – 4.17, p = 0.03) and multivariate (HR 2.35, 95% CI 1.12 – 4.92, p = 0.024) model.

Conclusions: SSRS is an effective strategy in achieving local control of spine metastases, particularly among radiosensitive histologies. However, a subset of these

ACCEPTED MANUSCRIPT

classically defined histologies (NSCLC and colorectal) has a propensity towards local failure. In addition to resulting in poorer OS outcomes, the poor LC rates seen in NSCLC and colorectal cancers in this study are more consistent with a radioresistant phenotype

RI PT

suggesting the need for optimized dosing regimens in this subgroup.

SC

Introduction

The skeletal system represents the third most common site of systemic metastasis with

M AN U

most cases involving the spine. Estimates have suggested that the incidence of metastasis to the spine is present in 5-30% of cancer patients (1, 2). Management of these spine metastases have involved radiotherapy as an effective treatment modality for establishing local control (LC), symptomatic management, and preventing spinal cord

TE D

compression (3).

Conventional external beam radiation therapy (EBRT) is effective in symptom palliation, but has limited success in LC for radioresistant spine tumors such as sarcomas and

EP

melanomas (4-6). As an alternative, spine stereotactic radiosurgery (SSRS) enables the delivery of ablative radiation doses while sparing nearby critical structures. Prospective

AC C

and retrospective studies have demonstrated effective LC rates of 72-96% with minimal toxicities (7). Yet, precise definitions of dosing schedules continue to evolve as we continue to understand the efficacy and safety in distinct tumor types (6, 8, 9). Our institutional practice involves treatment of radioresistant tumors to 24 Gy in 1 fraction, which achieves more effective long-term local control over lower dose regimens (10). In an effort to decrease risk of toxicities, radiosensitive tumors are treated to 27 Gy in 3 fractions or 18 Gy in 1 fraction, yet this dosing scheme may not be effective among all classically defined radiosensitive tumors.

ACCEPTED MANUSCRIPT

Although ambiguous in their classification, non-small cell lung cancer (NSCLC) and colorectal cancers are typically stratified as radiosensitive tumors even though they

RI PT

have historically presented with poorer overall survival (OS), progression free survival (PFS), and LC rates of metastases (11-16). This suggests that therapeutic intervention of these tumors may require strategies that are more applicable to radioresistant tumors

SC

such as sarcomas. Specifically, in cases of NSCLC and colorectal, the response rates after SSRS requires better characterization, since the assumption of a favorable

M AN U

treatment response like the one expected to other radiosensitive tumors such as breast cancer may not be consistently achieved. In such case, optimization of dosimetric strategies could lead to better outcomes. In this study, we present a cohort of retrospective and prospective patients treated with SSRS at our institution with metastatic lesions in the spine to investigate the clinical outcomes of NSCLC and

progression.

TE D

colorectal cancers with respect to clinical outcomes including OS, LC, and disease

AC C

Patients

EP

Materials and Methods

Following institutional review board approval, 127 patients with 148 spinal metastatic lesions treated with SSRS at XXXX between 2003 and 2013, were identified. Patients were followed for survival endpoints up until 2016. Of the 127 patients, 76 (59.8%), were treated on institutional Phase I/II clinical trials evaluating single and multifraction SSRS for spinal metastases (4, 17, 18). Radioresistant (i.e. melanoma, renal cell carcinoma, and sarcoma) and radiosensitive (i.e. thyroid, breast, colon, and NSCLC) classification was relative to conventional radiation and based on histology. Only radiosensitive

ACCEPTED MANUSCRIPT

tumors were included in this study. Epidural involvement was measured according to the epidural spinal cord compression scale (ESCC) (19). Burden of disease was classified exclusively based on the presence of a solitary metastatic spine lesion, multiple skeletal

RI PT

lesions, or disseminated visceral disease. Only patients treated with non-palliative SSRS doses were included in this study. Additionally, histologies with limited sample sizes

SC

were excluded.

Treatment

M AN U

Patients were treated with computed tomography (CT)-guided intensity modulated SSRS using the CT-on-rails EXaCT targeting system or the Truebeam 3 or Trilogy treatment delivery system with On-board Cone Beam CT (CBCT) (Varian Medical Systems as previously described (5, 20).

Patients were immobilized using an Elekta BodyFix

stereotactic body frame system (Elekta) and aligned using a CT-on-rails (GE Healthcare,

TE D

Fairfield, CT) or CBCT. Treatment planning was performed using intensity-modulated radiation therapy inverse-treatment planning software (Pinnacle, Philips Medical Systems). Treatments were monitored by the treating radiation oncologist and a

EP

dedicated radiation physicist to verify target positioning and quality assurance. Metastatic spinal tumors were treated with various dosing schedules based on

AC C

institutional protocols established and optimized for safety and efficacy as well as histological subtypes. Delineation of gross tumor volume (GTV) was defined based on MRI fused with the planning CT. The clinical tumor volume (CTV) contained the GTV along with contiguous bone marrow at risk or soft-tissue margin as defined by the consensus guidelines for target delineation (21). Planning treatment volume was defined with no margin to the CTV with the dose normalized between the 80% and 90% isodose line so that the prescription line enclosed the planning treatment volume, except near the spinal cord. As per institutional practice, dose and fractionation are not altered on the

ACCEPTED MANUSCRIPT

basis of tumor proximity to the spinal cord. Spinal cord constraints are prioritized over tumor coverage in treatment planning. Any steroid dosing given to patients at or near time of treatment is at the discretion of the treating physician. Patients who had been

RI PT

previously irradiated did not receive single fraction re-irradiation. Follow-up and End Points

Patients were evaluated during follow-up every 3 months with MRI of the spine and

SC

comprehensive physical and neurologic examination. Patterns of failure were evaluated by fusing SSRS treatment plans with the MRI dataset showing radiologic progression

M AN U

and determining recurrence location based on marginal (occurring between the 20% and 95% prescription isodose lines) and in-field (occurring entirely within the 95% isodose line) relapse subclassifications. For those treated off protocol, toxicity grading was retrospectively determined based on the observed and reported toxicities from each patient during treatment and at the time of follow-up using the Common Terminology

Statistical Analysis

TE D

Criteria for Adverse Events (CTCAE), version 4.0.

Baseline characteristics were evaluated using descriptive statistics. Survival analysis

EP

was calculated from the date of SSRS. LC was defined based on the time to local failure at the SSRS treated site, death was not classified as local failure. Patients were

AC C

censored at the date of last follow-up visit or death. Kaplan-Meier statistics were used for OS and LC rates. Log-rank tests were applied to determine differences between factors on survival curves. Cumulative incidence methodology was also used for local failure rates. Univariate and multivariate Cox proportional hazard regression models and competing-risk analyses were performed to risk factor assessment for OS and LC. The multivariate analysis for the Cox’s proportional hazard model and the competing-risk analyses was carried out using backwards elimination. All variables with a p-value of 0.25 or less on univariate analysis were included in the assessment with the least

ACCEPTED MANUSCRIPT

significant variable eliminated with each step until the most parsimonious model reported. The Wald test was used to assess the role of covariates in the model. This same method was used for the competing risk regression analysis. Survival outcomes

RI PT

for tumor type were stratified based on NSCLC and colorectal cancer status compared to other radiosensitive histologies.

SC

Results

M AN U

Patient Characteristics

Median duration of follow-up for all patients and for survivors was 22.6 months and 41.7 months respectively. Detailed patient, tumor, and treatment characteristics are listed in Table 1. The most common tumor histology was NSCLC (35.8%). Of the 127 patients treated, 17.3% and 9.4% had prior radiation or surgery to the site of SSRS, respectively.

TE D

No other sites of prior radiation or surgery were reported. Patients most commonly presented exclusively with disseminated metastatic spine lesions (44.9%), defined as two or more metastatic sites within the skeletal system. These tumors were most

EP

commonly found within the thoracic spine (57.4%) and 46.6% presenting with epidural disease. The majority of patients were women (59.1%), which is reflective of our breast

AC C

cancer patients being included in this study with a median age of 60 years (range 28-83 years). Several single fraction (SF) and multifraction (MF) SSRS regimens were used with the most common fractionation scheme being 27 Gy in 3 fractions (60.8%), followed by 18 Gy in 1 fraction (23%).

Overall Survival Univariate and multivariate Cox and competing risk analysis for OS are presented in Supplementary Table 1. Median OS time for all patients was 24.4 months with actuarial

ACCEPTED MANUSCRIPT

2-year OS rate of 51.5%. Tumor histology (NSCLC/colorectal cancer), ECOG performance status, male gender, and significant tumor burden were associated to poorer survival outcomes on Kaplan-Meier survival analysis (P =