Radiotherapy and Oncology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Radiotherapy and Oncology journal homepage: www.thegreenjournal.com

Blood–tumor barrier opening changes in brain metastases from pre to one-month post radiation therapy Feifei Teng a,d, Christina I. Tsien e, Theodore S. Lawrence a, Yue Cao a,b,c,⇑ a Department of Radiation Oncology; b Department of Radiology; c Department of Biomedical Engineering, University of Michigan, Ann Arbor, United States; d Department of Radiation Oncology, Shandong Cancer Hospital, Shandong University, Jinan, China; and e Department of Radiation Oncology, Washington University, St. Louis, United States

a r t i c l e

i n f o

Article history: Received 13 December 2016 Received in revised form 13 June 2017 Accepted 5 August 2017 Available online xxxx Keywords: Blood–tumor barrier Brain metastasis DCE MRI Radiotherapy

a b s t r a c t Purpose: Blood–tumor barrier is a limiting factor for effectiveness of systemic therapy to brain metastases. This study aimed to assess the extent and time course of BTB opening in BM following wholebrain radiotherapy (WBRT) or stereotactic radiosurgery (SRS) to determine optimal timing for systemic therapy. Materials and method: 30 patients received WBRT or SRS and a total of 64 metastatic lesions were analyzed. Dynamic contrast-enhanced MRI were acquired, to quantify a transfer constant (Ktrans), pre-RT, 1–2 weeks after starting RT (Wk1-2), and 1-month post-RT (1 M post-RT). Lesions were categorized as either low or high permeability based upon the pre-RT percentage volume of a lesion with Ktrans > 0.005 min 1 (%Vall) less or greater than 50%. Time-course changes of %Vall after RT were analyzed. Results: Fifty-seven lesions had high-permeability and seven had low-permeability at baseline. Intrapatient and inter-lesion heterogeneity was observed in six patients who had both low- (n = 7) and high-permeability lesions (n = 10). Also, lesion permeability showed a significant size-effect at baseline. For high-permeability lesions, either received WBRT (n = 43) or SRS (n = 14), %Vall decreased nonsignificantly following RT (from 85.4% pre-RT to 76.9% 1 M post-RT). For low-permeability lesions (n = 7, all received WBRT), %Vall increased from 5.6% pre-RT to 30.2% at Wk1-2 and to 52.6% 1 M post (p = 0.01). Conclusion: Our preliminary results suggest that 2–4 weeks after RT, when BTB opening is high for both low- and high-permeability brain metastatic lesions, could be optimal time to start systemic therapy. Ó 2017 Elsevier B.V. All rights reserved. Radiotherapy and Oncology xxx (2017) xxx–xxx

The blood–brain barrier (BBB) is composed of tight junctions arising between endothelial cells that as a whole act as a highly selective mechanism of permeability, regulating migration of various molecules between circulating blood and the brain. This structure protects the brain from harmful neurotoxins. Once cancer cells metastasize to the brain, however, the inherent selectivity of the BBB works against otherwise potentially effective systemic drug therapies [1]. As a result of the difficulty of drugs to cross the BBB, brain metastases is a potential ‘‘sanctuary” for disease that might otherwise be effectively treated with systemic therapies [2]. The BBB can be altered by disease and intervention. The presence of brain tumors can significantly affect or even disrupt brain endothelial junctions [3–5]. Tumor size may markedly impact the blood–tumor barrier (BTB) permeability, with larger tumors being more permeable and thereby more accessible to chemo-drug therapy and smaller metastases and micrometastases less so [6]. Radi⇑ Corresponding author at: Department of Radiation Oncology, University of Michigan, Ann Arbor, MI 48103, United States. E-mail address: [email protected] (Y. Cao).

ation has shown to increase the BTB permeability in animal models and patients [5,7,8], suggesting the potential enhancement of systemic anticancer therapy efficacy [8]. However, the optimal window for delivery of systemic therapy, presumably the timing for the maximum BTB opening, and the extent of BTB opening following RT remain largely unknown. The BTB permeability can be assessed quantitatively using dynamic contrast enhanced (DCE) MRI. The DCE-MRI can derive the transfer constant Ktrans, which describes the amount and the speed of accumulation of the gadolinium-based agent in the extra-vascular extra-cellular space through the BTB [9–11]. A pre-clinical study compares the disrupted BBB regions identified on the Ktrans map with Evans blue extravasation and indicates a match between the two [12]. Heterogeneous vascular leakage in brain metastases have been recognized [13,14]. Also, RT effects on high and low vessel leakage regions may be different. All these indicate that an analysis of the maximum, median or mean Ktrans within a tumor volume may be inadequate. This study aimed to analyze the time course and the extent of the BTB opening in brain metastases following whole-brain

http://dx.doi.org/10.1016/j.radonc.2017.08.006 0167-8140/Ó 2017 Elsevier B.V. All rights reserved.

Please cite this article in press as: Teng F et al. Blood–tumor barrier opening changes in brain metastases from pre to one-month post radiation therapy. Radiother Oncol (2017), http://dx.doi.org/10.1016/j.radonc.2017.08.006

2

Blood–tumor barrier opening changes in brain metastases from pre to one-month post radiation therapy

radiotherapy (WBRT) or stereotactic radiosurgery (SRS) by using Ktrans derived from DCE-MRI. Heterogeneous changes in the BTB permeability in low and high leaky metastatic lesions following RT were analyzed to determine the maximum timing window for the BTB opening. Methods Patients

A further evaluation showed widely distributed Ktrans values over 2–3 orders of magnitude within a single lesion, for which a mean value in a lesion does not well represent the characteristic Ktrans and changes. A cluster analysis identified Ktrans value of 0.037 min 1 to separate the high and low permeability regions in the lesions. Therefore, Ktrans > 0.037 min 1 was used to define a ‘‘high” BTB opening region within a lesion. The percentage GTVs having Ktrans > 0.037 min 1 defined as %Vhigh and Ktrans 0.005 min 1 as %Vlow at baseline and subsequent changes Wk1-2 and 1M-post were analyzed.

Thirty brain metastases patients participated in a prospective IRB approved MRI study and signed informed consents. Twentyone patients were treated with WBRT either 30 Gy in 3 Gy fractions or 37.5 Gy in 2.5 Gy/fraction. Nine patients were treated with SRS based on RTOG 93-05 using doses of 24 Gy, 18 Gy and 15 Gy. The analysis was limited to the three largest lesions in each patient or any lesion that was approximately 1 cc or greater. This resulted in a total of 64 lesions for analysis, 50 lesions from the 21 patients received WBRT and 14 lesions treated by SRS from the 9 patients.

Statistical Package for Social Sciences (IBM SPSS17.0) was used to determine significant differences in %Vall, %Vlow and %Vhigh following RT between low-permeability and high-permeability lesions by Student’s t-test. Further analyses were applied to each time points within each group using Student’s t-test. P values 0.5, supplementary Fig. 1). Therefore, we combined these patients in the analysis. First, we tried to understand the characteristics of lesion BTB permeability prior to RT. Of 64 evaluable lesions, the mean values (±SEM) of %Vall, %Vlow and %Vhigh pre-RT were 75.4% (±3.6%), 55.0% (±3.4%) and 20.4% (±3.1%), respectively, suggesting there was an extent of BTB opening at baseline. However, we did not find any significant differences in BTB opening related to histology or age using univariate analysis, (p > 0.05, Table 1). We did note a trend tumor-size effect on the BTB opening. For the small lesions 5 cc (n = 49), %Vlow was significantly greater but %Vhigh was marginally lower than the large tumors >5 cc (respective p-values of 0.025 and 0.055, Table 1). To further characterize lesion BTB opening at baseline, we categorized the lesions into high or low permeability based upon respective %Vall>or 0.05), neither received WBRT (43 lesions) nor SRS (14 lesions), see Fig. 2. For the low-permeability lesions (7 lesions, all treated by WBRT), a progressive increase in %Vall was observed from a baseline mean value (±SEM) of 6.4% (±3.1%) to 30.2% (±13.1%) Wk1-2 (p = 0.1), and significantly increased to 52.6% (±13.5%) (p = 0.01) 1 M postRT (Fig. 2). At the individual lesion level, of the 57 highpermeability lesions, nine lesions had a decrease in %Vall to 5 cc Concurrent chemo bortezomib No Radiation WBRT SRS *

Pts

Lesions

Mean %Vall

P value

Mean %Vlow

P value

Mean %Vhigh

P value

19 11

40 24

77.4% 72.1%

0.5

58.7% 48.8%

0.2

18.6% 23.3%

0.5

16 6 4 4

37 13 10 4

75.6% 76.3% 79.8% 59.1%

>0.05

55.3% 57.9% 51.9% 51.1%

>0.5

20.4% 18.4% 27.9% 8.0%

>0.05

19 11

27 37

75.6% 75.2%

0.9

55.1% 54.9%

1.0

20.5% 20.3%

1.0

27 10

49 15

75.3% 75.9%

1.0

59.3% 41.1%

0.025*

16.0% 34.8%

0.055

13 17

34 30

72.9% 78.3%

0.5

52.9% 57.5%

0.5

20.0% 20.8%

0.9

21 9

50 14

74.5% 78.6%

0.7

55.0% 55.0%

1.0

19.5% 23.5%

0.6

Statistically significant.

Fig. 1. Post-contrast T1-weighted images and Ktrans maps from a low-permeability tumor and a high-permeability tumor pre-RT, 2-weeks during WBRT and 1mo post-RT. Cyan, magenta, and red contours represent enhanced GTV, and the volumes with Ktrans > 0.005 and >0.037 min 1, respectively.

Fig. 2. Mean values (±SEM) of %Vall (a), %Vlow (b) and %Vhigh (c) of low-permeability and high-permeability tumors received WBRT and SRS pre-RT, 1–2 weeks after starting RT and 1 mo post-RT.

post-RT. Of the 7 low-permeability lesions, three had an increase in %Vall less than the group mean value as 0%, 27.5% and 29.5%. We also evaluated whether there were differences in the effect of radiation on %Vlow compared to %Vhigh between the low-

permeability and the high-permeability tumors. In the lowpermeability tumors, the changes in %Vlow showed a significant increase from a baseline value of 5.6% to 43.0%at 1 M-post (p = 0.01), indicating a radiation effect. There was a similar increase

Please cite this article in press as: Teng F et al. Blood–tumor barrier opening changes in brain metastases from pre to one-month post radiation therapy. Radiother Oncol (2017), http://dx.doi.org/10.1016/j.radonc.2017.08.006

4

Blood–tumor barrier opening changes in brain metastases from pre to one-month post radiation therapy

Fig. 3. Mean values (±SEM) of %Vall in low-permeability tumors (n = 7), and highpermeability tumors (n = 10) from the same 6 patients.

noted in the %Vhigh but to a smaller extent, from a baseline value of 0.8% to 9.6% 1 M-post. In the high-permeability tumors treated by either WBRT or SRS, %Vlow and %Vhigh did not show significant changes from baseline to 1 M-post (p > 0.05), see Fig. 2. We investigated if there was inter-lesion heterogeneity within a single patient. Six patients had at least one low-permeability lesion and one high-permeability lesion, indicating inter-lesion heterogeneity within the same patient. In the 6 patients, %Vall values of the low-permeability lesions (n = 7) were significantly different from those of the high-permeability tumors (n = 10) at baseline and at Wk1-2 of WBRT (p < 0.001 and 0.04, respectively), but no significant difference 1 M post-RT (Fig. 3), showing the vascular permeability heterogeneity of brain metastases within a single patient. Discussion In this study, we investigated the extent of the BTB opening of brain metastases pre-RT and the changes up to one-month postWBRT or SRS using a Ktrans derived from DCE-MRI. We found that the vascular permeability of brain metastases at baseline RT is very heterogeneous. Changes in vascular permeability in brain metastases following radiation are highly dependent on the baseline brain metastasis vascular permeability characteristics. Radiation does increase BTB permeability for the low-permeable metastatic lesions at baseline but changes the high-permeable lesions nonsubstantially. This heterogeneous radiation effect was also observed across multiple lesions within the same individuals. All these suggest that the previous view – radiation induces BTB disruption [8], is not always true. The marked heterogeneity of brain metastases as well as non-invasive MR imaging may provide an opportunity to consider a more personalized therapy. Optimizing the delivery of systemic agents to brain metastases may have an impact on response and local control [19,20]. Currently, the optimal timing of combining systemic therapies with WBRT is unclear [8]. Our study suggests that DCE-MRI can quantitatively assess the BTB opening in individual lesions over the course and after RT to guide on the selection of the optimal time, when the BTB opens to the maximum extent, to start systemic therapies. Our initial observations indicated that WBRT increased permeability in low permeability tumors and WBRT and SRS did not cause substantial changes in high permeability lesions up to one-month post-RT. Waiting 2–4 weeks after RT to start systematic therapy may benefit low permeability tumors for an increase in BTB opening but without a substantial decrease for high permeability lesions. Another important factor in optimizing the delivery of systemic agents in brain metastases is the extent of BTB opening. At the

baseline RT, most brain metastases have a mean Ktrans > 0.03 min 1. However, the permeability depends upon the molecular size of the contrast agent. Gd-DTPA is a relative small molecule (470 Da) compared to the size of many systemic agents. Therefore, elevated tumor vascular permeability based on Gd-DTPA may suggest sufficient permeability to enable ‘‘small molecular” drugs and tyrosine kinase inhibitors (TKIs) with molecular weights of 500 Da or less sufficiently cross the BTB to penetrate into brain metastases. However, large molecules such as antibodies up to 150 kDa [21–23], as well as the other properties of drugs, such as charge or specific efflux mechanisms, may also affect the ability of certain systemic agents to effectively cross the BTB. This particular area of research needs further investigation. In our study, the smaller tumors treated with SRS all demonstrated high leakage prior to RT. SRS is usually delivered with higher dose in one fraction, killing cells through apoptosis and giving little time for recovery and remodeling of the vascular endothelia [24]. A study evaluating large single doses of radiation on the cerebral vasculature reported that radiation increases the BBB permeability to molecules of various sizes [25]. Apoptosis of endothelial, glial cells, and neural, and neuro inflammation mediate radiation-induced secondary cell damage that leads to further disruption of BBB [5]. However, the increase in vascular permeability induced by a single fraction of high dose (20 Gy) happens as early as 24 h but can be delayed up to 90 days post-radiation [5]. Thus, it is possible that our observations were carried out outside of the time window for and that optimal BTB changes may occur beyond 1 month post SRS. We found a substantial heterogeneity in the tumor vascular permeability and the response following radiation. Radiationcaused BTB opening in low permeability lesions is most seen in the range of the Ktrans values between 0.005 and 0.037 min 1, which is relative low (Fig. 2b and c). The high-BTB opening in brain metastases at baseline is most likely caused by angiogenesis [26]. Another interesting finding in our study is the substantial intrapatient and inter-lesion heterogeneity prior to RT. The intrapatient and inter-lesion heterogeneity noted in our study suggests that a personalized approach may be needed in the treatment of brain metastases. In the future, specific treatment approaches to each individual brain metastasis may be required. We studied a group of patients with heterogeneous histologies, in which melanoma and renal cell cancer are considered to have high vascular leakage and high prevalent for bleeding at baseline. However, we did not observe any significant difference in the extent of BTB opening in melanoma lesions compared to NSCLC and breast cancer pre-RT (Table 1). Also, we did not find changes in BTB related histologies. Instead, we found the significant trend size differences in the vascular leakage at baseline as the small lesions (