- Email: [email protected]
Incidence of CNS Injury for a Cohort of 111 Patients Treated With Proton Therapy for Medulloblastoma: LET and RBE Associations for Areas of Injury
Accepted Manuscript Incidence of CNS Injury for a Cohort of 111 Patients Treated with Proton Therapy for Medulloblastoma; LET and RBE Associations for Areas of Injury Drosoula Giantsoudi, Roshan V. Sethi, Beow Y. Yeap, Bree R. Eaton, David H. Ebb, Paul A. Caruso, Otto Rapalino, Yen-Lin E. Chen, Judith Adams, Torunn I. Yock, Nancy J. Tarbell, Harald Paganetti, Shannon M. MacDonald PII:
S0360-3016(15)03343-X
DOI:
10.1016/j.ijrobp.2015.09.015
Reference:
ROB 23169
To appear in:
International Journal of Radiation Oncology • Biology • Physics
Received Date: 4 February 2015 Revised Date:
3 September 2015
Accepted Date: 11 September 2015
Please cite this article as: Giantsoudi D, Sethi RV, Yeap BY, Eaton BR, Ebb DH, Caruso PA, Rapalino O, Chen Y-LE, Adams J, Yock TI, Tarbell NJ, Paganetti H, MacDonald SM, Incidence of CNS Injury for a Cohort of 111 Patients Treated with Proton Therapy for Medulloblastoma; LET and RBE Associations for Areas of Injury, International Journal of Radiation Oncology • Biology • Physics (2015), doi: 10.1016/ j.ijrobp.2015.09.015. 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 Incidence of CNS Injury for a Cohort of 111 Patients Treated with Proton Therapy for Medulloblastoma; LET and RBE Associations for Areas of Injury. Drosoula Giantsoudi*, Roshan V. Sethi*, Beow Y. Yeap, Bree R. Eaton, David H. Ebb, Paul A. Caruso,
RI PT
Otto Rapalino, Yen-Lin E. Chen, Judith Adams, Torunn I. Yock, Nancy J. Tarbell, Harald Paganetti, Shannon M. MacDonald
Department of Radiation Oncology (D.G., R.V.S., B.R.E., Y.E.C., J.A., T.I.Y., N.J.T., H.P., S.M.M.), Department of Medicine (B.Y.Y.), Department of Pediatrics (D.H.E.) and Department of Radiology
SC
(P.A.C., O.R.) at the Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts, 02114 USA
Correspondence:
M AN U
Running Title: Correlations of LET and Brainstem toxicity after proton RT
Shannon MacDonald, M.D., Massachusetts General Hospital, Department of Radiation Oncology
Yawkey 112, 54 Fruit Street, Boston, MA 02114 Tel: (617) 726-5184, Fax: (617) 726-3603
TE D
E-mail: [email protected]
* DG and RVS has contributed equally to this work.
CONFLICT OF INTEREST NOTIFICATION: N.T.’s spouse is on the medical advisory board of ProCure.
EP
All other authors deny any real or potential conflicts of interest. FUNDING STATEMENT: R.V.S. was supported by the Doris Duke Charitable Foundation. Research was
AC C
supported by the National Cancer Institute of the National Institutes of Health under Award Number P01CA021239 and the Federal Share of program income earned by Massachusetts General Hospital on C06 CA059267, Proton Therapy Research and Treatment Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. ACKNOWLEDGMENTS: We would like to thank Jonathan Jackson and Tao Song of the Enterprise Research Infrastructure and Services (ERIS) group at Partners Healthcare for their in-depth support and smooth computing cluster operations, upgrades, and fixes.
1
ACCEPTED MANUSCRIPT
SUMMARY This report investigates the incidence of CNS injury in a large cohort of medulloblastoma patients treated with craniospinal irradiation by proton radiotherapy. An overall incidence of 3.6% was
RI PT
found. No clear correlation was found between the sites of toxicity and elevated RBE due to higher
AC C
EP
TE D
M AN U
SC
linear energy transfer values.
ACCEPTED MANUSCRIPT
ABSTRACT BACKGROUND: CNS injury is a rare complication of radiotherapy for pediatric brain tumors, but its incidence with proton radiotherapy (PRT) is less well defined. Increased linear energy
RI PT
transfer (LET) and relative biological effectiveness (RBE) at the distal end of proton beams may influence this risk. We report incidence of CNS injury in medulloblastoma patients treated with PRT and investigate correlations with LET and RBE values.
SC
METHODS: We reviewed 111 consecutive patients treated with PRT for medulloblastoma between 2002 and 2011 and selected patients with clinical symptoms of CNS injury. MRI
M AN U
findings for all patients were contoured on original planning scans (TCAs-treatment change areas). Dose and LET distributions were calculated for the treated plans using Monte Carlo. RBE values were estimated based on LET-based published models.
RESULTS: At 4.2 years median follow up, the 5-year cumulative incidence of CNS injury was
TE D
3.6% for any grade and 2.7% for grade 3+. Three of four symptomatic patients were treated with a whole posterior fossa boost. Eight of ten defined TCAs had higher LET values than the target but statistically non-significant differences in RBE values (p-value=0.12).
EP
CONCLUSIONS: CNS/brainstem injury incidence for PRT in this series is similar to that reported for photon radiotherapy. The risk of CNS injury was higher for whole posterior fossa
AC C
boost than for involved field. Although no clear correlation with RBE values was found, numbers were small and additional investigation is warranted to better determine the relationship between injury and LET.
Key Words: Medulloblastoma, proton, brainstem injury, CNS injury, linear energy transference
1
ACCEPTED MANUSCRIPT
INTRODUCTION As medulloblastoma outcomes have improved markedly, long-term toxicities in survivors are receiving greater attention. CNS radiation injury, most often seen in the brainstem, is a rare
RI PT
but potentially devastating treatment complication(1). PRT is becoming more widely used for medulloblastoma due to better sparing critical CNS structures and organs anterior to the brainstem and spinal column and its potential to decrease late complications. However, at least a portion
SC
and sometimes the entire brainstem is part of the high-dose target volume for PRT and ideal beam arrangement for sparing the cochlea, neuroendocrine structures and temporal lobes usually
M AN U
requires placing the distal end of the treatment field within the brainstem for an involved field (IF) boost and at or just beyond it for whole posterior fossa (PF) treatment. For PRT, it’s necessary to account for the relative biological effectiveness (RBE) compared to photon treatments. Although an average RBE value of 1.1 is used clinically, it’s
TE D
known that RBE varies within a PRT field, depending dose, linear energy transfer (LET) and radiobiological properties of tissues(2). LET is the energy transferred to absorbing tissue per unit track length of a particle reported in keV/µm and maximizes near the end of proton field range.
EP
LET could impact the incidence of CNS injury from PRT if high LET regions are within the brainstem and RBE-weighted-doses are higher than estimated by current treatment planning
AC C
systems. Clinical experience is maturing and software systems have been developed to accurately calculate the LET distribution for a proton plan within a patient facilitating the investigation of potential correlation of clinical outcomes with LET distributions, to further our knowledge of PRT and potentially improve upon treatment planning for future patients. In this study, we review the incidence and location of CNS and brainstem injury in the largest existing medulloblastoma patient cohort treated with PRT. We calculated the LET
2
ACCEPTED MANUSCRIPT
distribution for patients with and without injury and compared LET values of MRI treatment change areas (TCAs) to those for the whole treated volumes.
RI PT
MATERIALS AND METHODS
Patients and clinical data
SC
Between May 2002 and December 2011, 111 pediatric patients with medulloblastoma were treated with CSI PRT at our institution. Eighty-four of these patients were enrolled on
M AN U
prospective institutional trials. Clinical data obtained from charts review included the patient’s sex, ethnicity, date of birth, pathological diagnosis, histological grade, risk status, surgical procedures, extent of surgery, PRT treatment details, neurological symptoms, date of last follow-
TE D
up, disease and survival status, location and date of first observed symptomatic TCA.
Definition of radiation injury
We defined radiation injury as new or progressive CNS symptoms not attributable to
EP
tumor progression. We included all areas of MRI radiographic changes in symptomatic patients. Consistent with previous publications, injury was retrospectively graded using the National
AC C
Cancer Institute Common Terminology Criteria for Adverse Events (version 4) for central nervous system necrosis(1, 3). (See appendix). Not all cases were considered to represent necrosis, which is why the term “injury” is used preferentially in this manuscript. Areas of MRI radiographic changes in asymptomatic patients included in this study were not reported as radiation injury, since they were not associated with any CNS symptoms.
LET and RBE-weighted-dose distribution 3
ACCEPTED MANUSCRIPT
CSI passively scattered PRT was delivered to all patients in the prone position, followed by either an IF boost or whole PF boost. For patients with reported radiographic changes (TCAs), MRI scans were fused with the original planning CT image set. The locations of radiographic
RI PT
changes were then contoured in the original CT by a neuro-radiologist (P.C.) and a pediatric radiation oncologist (S.M.) based on registration and anatomic landmarks and accounting for growth and change.
SC
Each patient’s anatomical and beam geometry were imported, as treated, into TOPAS (TOol for PArticle Simulation – version b8(4)) Monte Carlo system. Using the clinical dose
M AN U
calculation grid, dose-to-water and corresponding dose-averaged LET (LETd) values, in units of keV/µm/(gr/cm3), were scored in voxel-by-voxel basis, as described by Grassberger et al.(5). To evaluate potential clinical implications associated with the LET distribution, voxel-byvoxel RBE and RBE-weighted-doses were calculated for all patients with TCAs and six
TE D
additional randomly selected patients without reported TCAs. Because RBE estimations are associated with considerable uncertainties and variations among different models, two alternative LET-based RBE models were employed here, described by Freese et al(6) (“Wilkens model”) and
EP
Carabe et al(7) (“Carabe model”), respectively. Both models are based on the standard linearquadratic dose-response formalism and assume linear relationship between LET and RBE for a
AC C
given dose, with the slope depending on the α/β ratio of the tissue. In the Wilkens model, an average target RBE value is considered as an input parameter
along with its mean dose and LET values calculated based on the clinical treatment plan information. The model predicts the relative variations in RBE values, while maintaining the target’s average RBE value. In this work, an average target RBE value of 1.1 was assumed as an input. In contrast, the Carabe model predicts absolute RBE values based on the calculated dose and LET distributions for each structure of reference. Differences in tissue response are 4
ACCEPTED MANUSCRIPT
considered via α/β values in both models. For the structures (i.e. normal brain, healthy CNS tissues) under study an α/β value of 2.1 Gy was used in both models (αx=0.0425 Gy-1 and
RI PT
βx=0.0203 Gy-2), as averaged over the range of values found in literature(8, 9).
Statistics
Time to injury was measured from the starting date of PRT until the date of first observed TCA or
SC
was censored at the last follow-up among patients still alive without injury. The cumulative incidence of radiation injury was estimated with death defined as a competing risk and was
M AN U
compared between dose groups using Gray’s test. LET and RBE value comparisons between the necrotic areas or TCAs and the whole boost target volume, were assessed using the Wilcoxon signed-rank test. Data analysis was performed using SAS 9.3 (SAS Inst Inc, Cary, NC) with p-
RESULTS Patients
TE D
values based on a two-sided hypothesis test.
EP
Table 1 summarizes the demographic, tumor and treatment characteristics of all patients with (n=4) and without (n=107) CNS radiation injury. Also listed are those patients who showed
AC C
radiographic treatment change but developed no symptoms (6 out of the 107). The majority of patients with standard risk disease were treated per or on ACNS0331 and the majority of high-risk patients were treated per or on ACNS0332. A small number of younger patients were treated on or per Headstart II or III protocol prior to radiation. Nearly all patients presented with hydrocephalus. Most had resolution of hydrocephalus with removal of tumor and few required a permanent shunt. None of the 4 patients with CNS injury required a shunt. All patients with symptoms of injury had radiographic findings. No patient underwent biopsy. Clinically, follow 5
ACCEPTED MANUSCRIPT
up was considered sufficient to ensure that these were not cases of tumor recurrence. All patients had areas of gadolinium enhancement and larger areas of T2 hyperintensity.
RI PT
[Table 1]
Radiation injury
Ten patients had post-treatment radiographic changes, of whom four patients experienced
SC
CNS radiation injury at a median time of 9 months (range, 8 to 18 months) from the start of PRT. Table 2 summarizes the clinical details of the four symptomatic patients. Patients 1,3, and 4 were
M AN U
treated with chemotherapy on or per ACNS0331 and patient 2 received a modified Head Start II regimen. This patient received 2 cycles of Methotrexate, Cisplatin, Cytoxin, and Etopside. He developed acute renal failure and chemotherapy was modifed to include oral Etoposide, Vincristine and Carboplatin. Three patients developed brainstem injury symptoms with All three patients received
TE D
corresponding MRI lesions (two grade 3 injury; one grade 4).
treatment. Two recovered fully from their symptoms, but the patient with grade 4 injury remains paraplegic and dependent on a tracheostomy and feeding tube. The fourth patient developed
EP
cervical injury symptoms and a corresponding lesion was found at C1-C2, considered as grade 2 injury. This patient received no intervention other than omission of her final cycle of
AC C
chemotherapy and recovered completely. Two patients developed additional symptomatic radiation injury more than a month after the initial injury. Patient 1 developed osteonecrosis of the right temporal bone (internal auditory canal) 16 months after the first injury. In addition, this patient developed an additional lesion consistent with brainstem necrosis 6 years and 8 months after the first injury. Patient 3 developed additional brainstem injury 27.4 months from the start of RT.
6
ACCEPTED MANUSCRIPT
[Table 2]
The 5-year cumulative incidence of CNS radiation injury was 3.6% for grade 2-4 injury,
RI PT
and 2.7% for with Grade 3+ injury (see Table 3). As all patients with brainstem lesions developed Grade 3+ brainstem injury, the 5-year cumulative incidence of brainstem radiation injury/necrosis is also 2.7%. The risk of all or Grade 3+ only radiation injury was not significantly altered by CSI
SC
dose (>23.4 Gy(RBE) vs. < 23.4 Gy(RBE): 0% vs. 5.3%; p=0.179 and 0% vs. 4.0%; p=0.247) or boost dose (<54.0 Gy(RBE) vs. >54.0 Gy(RBE): 3.8% vs. 0%; p=0.693 and 2.9% vs. 0%;
M AN U
p=0.733). The risk of any radiation injury was higher with the use of a whole PF boost (IF boost vs. whole PF boost; 1.4% vs. 7.7%; p=0.094) but the difference in only Grade 3+ injury was not statistically significant (1.4% vs. 5.1%; p=0.259).
TE D
[Table 3]
Studying the suggested risk factors per Indelicato et al(3), for the three patients who
EP
experienced brainstem injury, no brainstem volume received dose equal or higher than 60 Gy (Table 2). However for three patients (patients 1, 3 and 4; all received PF boost) the median dose
AC C
to the brainstem was higher than 52.4 Gy(3). The same was true for all 6 patients with asymptomatic TCAs, although none of the TCAs in these cases were located within the brainstem.
LET and RBE-weighted-dose calculations Table 4 summarizes the simulated mean LET and RBE-weighted-dose values for all patients under study. The mean LET value for the total of 10 symptomatic TCAs is compared to 7
ACCEPTED MANUSCRIPT
the corresponding value for the whole boost target volume, which was the whole PF for three patients and the involved field for one. Eight out of ten symptomatic TCAs showed higher LET values compared to the whole target (range of differences: 0.02-1.4 keV/µm per unit density).
RI PT
The mean LET value, averaged over all 10 symptomatic TCAs, was 2.7 versus 2.4 keV/µm per unit density in the whole boost target volume. Differences in LET values between TCAs and the corresponding whole boost target volume were statistically significant (p-value<0.05).
SC
The mean LET value, averaged over all 6 asymptomatic TCAs, was lower than the corresponding value for the boost target volumes (2.2 versus 2.5 keV/µm/(gr/cm3), p-value<0.05)
M AN U
but the difference in biological effect between these two LET values would be insignificant (~2%). No statistically significant differences were found on the volume of brainstem with LET more that 3keV/µm/(gr/cm3) between the symptomatic, the asymptomatic and the no-TCA patients. No differences were observed in the LET distributions among the three studied patient
TE D
groups (Appendix figures 1-5).
EP
[Table 4]
For all symptomatic patients the Carabe model predicted higher mean RBE-weighted-dose
AC C
for the boost target and TCAs compared to the constant RBE value of 1.1. Figures 1 and 2 show the results for patients 1 and 2 (tables 2 and 4). Two-dimensional RBE-weighted-dose distributions assuming constant (figures 1a and 2a) and variable RBE (Carabe model - figure 1b and 2b) are presented, along with their difference (figures 1c-d and 2c-d). For patient 1, great variations in LET distributions were noticed within the PF target and among the injured areas, as seen by the distribution of the LET volume histograms (LET-VHs - figure 1e) and the mean values in table 4. The RBE-weighted DVHs for these areas followed a similar pattern with the 8
ACCEPTED MANUSCRIPT
LET-VHs: although for constant RBE their mean doses are close or even lower than the mean target dose, all TCAs, except for the resection cavity TCA, showed increased variable-RBE-
[Figure 1]
RI PT
weighted dose compared to the whole target volume.
SC
Similarly for patient 2 (figure 2), the TCA’s mean RBE-weighted-dose (Carabe model) was found to be 3.1 Gy higher than the corresponding value for the target (figures 1c and d and
M AN U
table 4). The LET-VHs for this case (figure 2e) show considerably increased LET distribution for the TCA (brainstem SRC – figure 2) compared to the IF target (CTV). DVH analysis shows higher variable-RBE-weighted-doses for the brainstem TCA for this case compared to the CTV, while for constant RBE the inverse is observed (figure 2f). Color-wash areas in Figure 2f
TE D
represent the uncertainty due to the range of α/β values reported in the literature for the structures of interest (α/β ranging from 1.5 Gy to 2.8 Gy for normal brain anatomy(5)).
EP
[Figure 2]
AC C
Based on the reported values (table 4 - Carabe model), mean RBE averaged over all 10
symptomatic TCAs was 2% higher compared to the corresponding target mean RBE, but not statistically significant (p=0.12). Furthermore, 2% dose deviations are well within the accuracy of clinical dosimetry and considered clinically insignificant. Similar results were observed with the Wilkens model: under the assumption of average target RBE of 1.1, the mean RBE-weighteddose for the TCAs was not statistically different than when assuming constant RBE.
9
ACCEPTED MANUSCRIPT
DISCUSSION: Our five-year cumulative incidence of overall CNS injury is 3.6%. The cumulative incidence of >Grade 2 injury or brainstem injury is 2.7%. Initial affected sites included the
RI PT
brainstem (n=3) and upper cervical spinal cord (n=1) for this cohort.
The incidence of brainstem radiation injury/necrosis encountered in this series is comparable to photon series, which are summarized in the appendix(1, 10-13). In the largest
SC
retrospective review, Murphy et al followed 236 children with embryonal tumors for a median of 52 months(1). The 5-year cumulative incidence of any radiation necrosis was 3.7% among all
M AN U
patients, and 4.4% among those with infratentorial tumors. The second largest series, by Christopherson et al, followed 53 patients with medulloblastoma for a median of 24 years, with a crude incidence of symptomatic radiation necrosis of 5.6%(10). Although these figures are similar to those encountered in our series, there are several limitations to the comparison. In the
TE D
study by Murphy et al, the primary outcome was imaging changes, though seven of eight affected patients were symptomatic. In addition, Murphy et al administered a higher median IF or PF boost dose compared to our series and a relatively aggressive chemotherapeutic regimen(1).
EP
Similar to Murphy, we found that after the first diagnosis of injury or necrosis, subsequent injury can still develop, even after resolution of the first event.
AC C
There is little available data on CNS injury for PRT(14, 15). Indelicato et al reviewed
children treated with PRT for a variety of central nervous system malignancies(3). Among the 313 patients who received a minimum dose of 50.4 Gy(RBE) to the brainstem, the 2-year cumulative incidence of any brainstem injury was 3.8% and of grade 3+ injury was 2.1%. Rates of injury were higher for PF tumors. MacDonald et al followed 70 patients with ependymoma treated with PRT to a median dose of 55.8 Gy(RBE) to areas at high risk for disease recurrence;
10
ACCEPTED MANUSCRIPT
the majority of patients received 54 Gy(RBE) to the initial CTV(14, 15). At a median follow-up of 46 months, there were no cases of radiation necrosis. It is worth noting that Ruben et al found that sensitizing chemotherapy regimens increased
RI PT
the risk of radiation necrosis by fivefold(16). Spearfico et al reported the highest recorded incidence of symptomatic radiation necrosis in a study of children who received myeloablative chemotherapy (followed by peripheral blood stem cell rescue) in addition to photon RT(13). In
SC
our series all patients with grade 3 brainstem injury/necrosis had PF syndrome and notable sensitivities to chemotherapeutic agents, with one patient tested for DNA sensitivity due to severe
M AN U
chemotherapy injury prior to receiving PRT. While this is only an observation it supports the hypothesis that multiple patient and treatment-related factors contribute to the etiology of brainstem necrosis. The influence of other variables, including dose, gross total resection, perioperative complications and shunting, should also not be discounted. In our series, no significant
TE D
effect was found from the CSI dose or the cumulative involved field dose. We did find an association with boost treatment of the whole PF versus the IF. Due to the limited number of symptomatic patients in our series, it is difficult to draw firm conclusions.
EP
However, this finding supports Murphy’s findings that patients who developed necrosis received >50Gy to a larger percentage of the infratentorial brain (92.1% vs. 72.9%; p=0.03)(1) and
AC C
Indelicato’s finding that median dose to the brainstem of >52.4 Gy is a risk factor for brainstem injury. Some centers employ techniques to avoid placing the end of range in the brainstem by extending the field’s range beyond the brainstem(17). Apart from increasing the brainstem volume receiving full dose, these techniques increase the dose to critical structures beyond the brainstem (cochlea, neuroendocrine structures and temporal lobes). Caution is advised in using them as our findings along with Murphy’s and Indelicato’s suggest they may increase the risk of injury. 11
ACCEPTED MANUSCRIPT
Although our statistical sample of symptomatic TCAs was relatively small, the differences in mean RBE values and RBE-weighted-doses, among the three patient groups that were analyzed, were not statistically or clinically significant. Considering the uncertainties of the RBE
RI PT
models, due to reported range of α/β ratios for the structures of interest, significant variability in potential RBE-weighted-dose distributions is possible (figure 1f). Although variable-RBEweighted DVHs show higher organ doses than those assuming constant RBE value, overlapping
SC
areas of the DVH bands (blue and red) in figure 1f suggests significant uncertainty in the relative RBE-weighted dose difference between TCAs and whole target volumes.
M AN U
In summary, no current evidence exists supporting higher incidence of CNS radiation injury in medulloblastoma patients treated with PRT compared to photon RT. We found higher rate of brainstem injury/necrosis for medullobastoma patients compared to ependymoma patients treated with protons despite comparable doses to similar region. However, we should recognize
TE D
that radiation doses for both these patient cohorts were relatively conservative, rarely exceeding a prescription dose of 54 Gy(RBE) to the CTV and making every effort to avoid brainstem doses exceeding 54 Gy(RBE). Although LET levels were higher in symptomatic TCAs, no clear
EP
correlation was found between the sites of injury and elevated RBE due to higher LET values. Brainstem LET distributions were similar in all patients under study. Furthermore, the greater
AC C
likelihood of brainstem injury in treatment of PF versus IF indicates that the treated volume may play a more significant role for these cases. PRT is a promising treatment for children, with known physical properties that spare
healthy tissues from high and low radiation doses. When administering conservative doses, brainstem injury rates were comparable with the photon experience.
Future directions may
include testing for additional patient related risk factors for radiation injury, such as DNA
12
ACCEPTED MANUSCRIPT
sensitivity or other biomarkers. Further study of clinical outcomes to better explore biological
RI PT
endpoints is still needed.
REFERENCES
SC
1. Murphy ES, Merchant TE, Wu S, et al. Necrosis after craniospinal irradiation: results from a prospective series of children with central nervous system embryonal tumors. Int. J. Radiat. Oncol. Biol. Phys. 2012;83:e655–60.
M AN U
2. Paganetti H, Niemierko A, Ancukiewicz M, et al. Relative biological effectiveness (RBE) values for proton beam therapy. International Journal of Radiation Oncology*Biology*Physics. 2002;53:407–421. 3. Indelicato DJ, Flampouri S, Rotondo RL, et al. Incidence and dosimetric parameters of pediatric brainstem toxicity following proton therapy. Acta Oncol. 2014:1–7. 4. Perl J, Shin J, Schümann J, et al. TOPAS: An innovative proton Monte Carlo platform for research and clinical applications. Med. Phys. 2012;39:6818–6837.
TE D
5. Grassberger C, Trofimov A, Lomax A, et al. Variations in Linear Energy Transfer Within Clinical Proton Therapy Fields and the Potential for Biological Treatment Planning. International Journal of Radiation Oncology*Biology*Physics. 2011;80:1559–1566.
EP
6. Frese MC, Wilkens JJ, Huber PE, et al. Application of constant vs. variable relative biological effectiveness in treatment planning of intensity-modulated proton therapy. Int. J. Radiat. Oncol. Biol. Phys. 2011;79:80–88. 7. Carabe A, Moteabbed M, Depauw N, et al. Range uncertainty in proton therapy due to variable biological effectiveness. Phys. Med. Biol. 2012;57:1159–1172.
AC C
8. Meeks SL, Buatti JM, Foote KD, et al. Calculation of cranial nerve complication probability for acoustic neuroma radiosurgery. International Journal of Radiation Oncology*Biology*Physics. 2000;47:597–602. 9. Hornsey S, Morris CC, Myers R, et al. Relative biological effectiveness for damage to the central nervous system by neutrons. International Journal of Radiation Oncology*Biology*Physics. 1981;7:185–189. 10. Christopherson KM, Rotondo RL, Bradley JA, et al. Late toxicity following craniospinal radiation for early-stage medulloblastoma. Acta Oncol. 2014;53:471–480. 11. Fouladi M, Chintagumpala M, Laningham FH, et al. White matter lesions detected by magnetic resonance imaging after radiotherapy and high-dose chemotherapy in children with medulloblastoma or primitive neuroectodermal tumor. J. Clin. Oncol. 2004;22:4551–4560. 13
ACCEPTED MANUSCRIPT
12. Merchant TE, Li C, Xiong X, et al. Conformal radiotherapy after surgery for paediatric ependymoma: a prospective study. The Lancet Oncology. 2009;10:258–266. 13. Spreafico F, Gandola L, Marchianò A. Brain Magnetic Resonance Imaging After High-Dose Chemotherapy and Radiotherapy for Childhood Brain Tumors. International Journal of …. 2008.
RI PT
14. MacDonald SM, Safai S, Trofimov A, et al. Proton radiotherapy for childhood ependymoma: initial clinical outcomes and dose comparisons. Int. J. Radiat. Oncol. Biol. Phys. 2008;71:979– 986. 15. MacDonald SM, Sethi R, Lavally B, et al. Proton radiotherapy for pediatric central nervous system ependymoma: clinical outcomes for 70 patients. Neuro-oncology. 2013;15:1552–1559.
SC
16. Ruben JD, Dally M, Bailey M, et al. Cerebral radiation necrosis: incidence, outcomes, and risk factors with emphasis on radiation parameters and chemotherapy. Int. J. Radiat. Oncol. Biol. Phys. 2006;65:499–508.
AC C
EP
TE D
M AN U
17. Buchsbaum JC, McDonald MW, Johnstone PA, et al. Range modulation in proton therapy planning: a simple method for mitigating effects of increased relative biological effectiveness at the end-of-range of clinical proton beams. Radiat Oncol. 2014;9:2.
14
ACCEPTED MANUSCRIPT
FIGURE CAPTIONS Figure 1: (Patient 1) RBE-weighted-dose for constant(a) and variable RBE(b), axial(c) and
RI PT
sagittal(d) views of their difference, LET-VHs(e) and RBE-weighted DVHs(f). PF: posterior fossa; TCA: treatment change area; RBEw: RBE-weighted; IAC: Inner Acoustic
SC
Canal;
M AN U
Figure 2: (Patient 2) RBE-weighted-dose for constant(a) and variable RBE(b), axial(c) and sagittal(d) views of their difference, LET-VHs(e) and RBE-weighted DVHs(f) for constant (thick dashed lines) and variable RBE (Carabe model) for a range of a/b ratios (thin solid lines: α/β=2.8 Gy; thick solid lines: α/β=2.1 Gy; thin dashed lines: α/β=1.5 Gy).
AC C
EP
TE D
CTV: clinical target volume (boost); TCA: treatment change area; RBEw: RBE-weighted
15
ACCEPTED MANUSCRIPT
10.5 years 6 years - 22 years 0 (0%) 4 (100%)
66 (59%) 45 (41%)
2 (50%) 2 (50%)
7 years 32 months - 22 years 2 (2%) 105 (98%)
2 (33%) 4 (67%)
64 (60%) 43 (40%)
4 (100%) 0 (0%) 0 (0%) 0 (0%)
6 (100%) 0 (0%) 0 (0%) 0 (0%)
92 (86%) 2 (2%) 6 (6%) 7 (6%)
4 (100%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
5 (83%) 1 (17%) 0 (0%) 0 (0%) 0 (0%)
77 (72%) 18 (17%) 10 (9%) 1 (1%) 1 (1%)
76 (68%) 35 (32%)
4 (100%) 0 (0%)
5 (83%) 1 (17%)
72 (67%) 35 (33%)
5 (4%) 1 (1%) 71 (64%) 3 (3%) 2 (2%) 1 (1%) 28 (25%)
0 (0%) 0 (0%) 4 (100%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
0 (0%) 1 (17%) 3 (50%) 0 (0%) 0 (0%) 0 (0%) 2 (33%)
5 (5%) 1 (1%) 67 (63%) 3 (3%) 2 (2%) 1 (1%) 28 (26%)
0 (0%) 0 (0%) 4 (100%) 0 (0%) 0 (0%)
0 (0%) 0 (0%) 5 (83%) 1 (17%) 0 (0%)
3 (3%) 1 (1%) 99 (93%) 3 (3%) 1 (1%)
96 (86%) 2 (2%) 6 (5%) 7 (6%)
TE D
81 (73%) 18 (16%) 10 (9%) 1 (1%) 1 (1%)
AC C
EP
Anaplastic+Desmoplastic Nodular Risk Classification Standard High CSI PRT Dose (Gy[RBE]) 18 22.5 23.4 27 30.6 34.2 36 Boost Field PRT Dose (Gy[RBE]) 50.4 52.2 54.0 55.8 59.4
Patients without injury (including patients with asymptomatic treatment change)(n=107)
8 years 27 months – 14 years 1 (17%) 5 (83%)
3 (3%) 1 (1%) 103 (93%) 3 (3%) 1 (1%)
SC
Gender Male Female Race Caucasian African-American Asian Other/Unknown Histology Classic Anaplastic Desmoplastic
7 years 32 months - 22 years 2 (2%) 109 (98%)
Patients with asymptomatic radiographic treatment change (n=6)
M AN U
Age at Treatment Median Range < 3 years > 3 years
Patients with injury who developed symptoms (n=4)
RI PT
Table 1: Patient characteristics All patients (n=111)
1
ACCEPTED MANUSCRIPT
69 (62%) 42 (38%) 23.4 (5.4 – 36.0)
1 (25%) 3 (75%) (30.6 – 30.6)
3 (50%) 3 (50%) (18.0 – 31.5)
67 (63%) 40 (37%) 19.8 (5.4 – 36.0)
AC C
EP
TE D
M AN U
SC
RI PT
Boost (Gy[RBE]) Involved Field only Whole Posterior Fossa Median (range)
2
ACCEPTED MANUSCRIPT
8.2/ F
SR
23.4
WPF (30.6)
55.00
56.17
7.9
Leg weakness, ataxia
23.4
3
23.0/M
SR
23.4
WPF (30.6)
5
13.4/F
SR
23.4
WPF (30.6)
83
Dysarthria, ataxia
Brainstem
Steroids, Avastin
Brainstem, upper cervical cord
3
HBOT
AliveNED,NRS
Brainstem
4
Steroids, Avastin
AliveAWD,RS
55.00
9.0
54.55
55.85
17.8
Left hemiparesis
54.56
27.4 56.02
AliveNED,RS
None
50.17
AC C
4
Steroids, HBOT
24
M AN U
SR
TE D
6.4/M
3
Right-sided hearing loss*
Right-sided weakness
EP
2
IF (30.6)
Brainstem, cerebellum
Patient Current Status
Right temporal bone (internal auditory canal)
SC
1
Symptoms
Location of Treatment MRI Grade and changes intervention
RI PT
Table 2: Clinical characteristics of patients with injury Whole Patient PF / Median Maximum Time to Patient age Risk CSI Dose IF(D50) Dose to injury ID (y)/ Group [Gy(RBE)] only Dose to Brainstem imaging Sex (Dose Brainstem [Gy(RBE)] (mo) in Gy)
8.9
Weakness
Brainstem, cerebellum Upper cervical cord
Steroids 2
None
AliveNED,NRS
SR, standard risk; HBOT, hyperbaric oxygen therapy; IF, involved-field only boost; WPF, whole posterior fossa boost. Note: patient age is the age at which the patient underwent radiation; NED: No evidence of disease; AWD: Alive with disease; NRS: No residual symptoms; RS: Residual symptoms of toxicity. *Hearing loss developed before imaging evidence of osteoradionecrosis and may be due to Cisplatin and/or cochlear dose from RT
3
ACCEPTED MANUSCRIPT
PF: posterior fossa
AC C
EP
TE D
M AN U
10
SC
RI PT
Table 3: Crude and cumulative incidence of symptomatic necrosis Grades 2-4 Grade 3+ only Crude 5-year p-value Crude 5-year p-value rate incidence rate incidence All patients 4% 4% 3% 3% CSI Dose (Gy[RBE]) 0.179 x >23.4 0% 0% 0% 0% <23.4 4% 5% 3% 4% Whole PF Boost 0.612 0.862 Involved Field only 2% 3% 2% 3% Whole PF Boost 2% 5% 1% 2% Involved Field (Gy[RBE]) 0.693 0.733 < 54.0 4% 4% 3% 3% > 54.0 0% 0% 0% 0%
4
ACCEPTED MANUSCRIPT Table 4: LET, RBE and RBE-weighted (DRBE) dose calculations results
3
4
AC C
2
EP
TE D
M AN U
SC
1
Mean Dose (CARAB E) a/b = 2.1 [Gy(RB E)]
RI PT
Patie nt
Mean Dose Mean (WILKE Mean Mean LET Dose NS) V(LET> RBE Structur [keV/µm/(u (RBE = a=0.042 3) (WILKE e nit 1.1) 5, [%] NS) density)] [Gy(RB b=0.020 E)] 3 [Gy(RBE )] Symptomatic Treatment Change (Injury) PF 2.36 5.8% 51.0 51.0 1.10 Target Brainste 2.77 20.2% 50.5 51.4 1.12 m Cerebell 2.40 0.1% 51.2 51.3 1.10 um STC Inner Acoustic 3.26 77.3% 50.2 52.4 1.15 Canal STC Brainste 3.01 47.4% 50.6 52.2 1.13 m STC 1 Brainste 2.38 0.0% 51.1 51.2 1.10 m STC 2 Resectio n Cavity 2.03 0.0% 51.3 50.5 1.08 STC Brainste 2.51 0.0% 51.0 51.4 1.11 m STC 3 PF 2.61 14.8% 54.1 54.1 1.10 Target Brainste 3.91 80.6% 45.1 48.0 1.17 m Brainste 4.01 96.5% 52.5 55.7 1.17 m STC PF 2.34 3.1% 54.8 54.8 1.10 Target Brainste 3.17 43.8% 52.7 54.9 1.14 m Brainste 2.75 1.9% 54.7 55.8 1.12 m STC Cerebell 2.24 0.0% 55.0 54.8 1.09 um STC PF 2.24 0.9% 54.4 54.4 1.10 Target
Mean RBE (CARAB E)
54.7
1.18
55.5
1.21
55.0
1.18
56.8
1.25
56.4
1.23
54.9
1.18
53.9
1.15
55.2
1.19
59.0
1.20
53.5
1.31
62.1
1.30
58.7
1.18
59.3
1.24
60.1
1.21
58.6
1.17
57.9
1.17 5
ACCEPTED MANUSCRIPT Brainste m Cervical Spine STC
2.47
2.1%
54.3
54.9
1.11
58.6
1.19
2.80
0.0%
38.5
40.5
1.16
43.1
1.23
1.10
58.9
1.20
9
10
11
12
13
54.1
54.1
3.32
51.4%
52.6
54.2
2.20
0.0%
55.1
54.1
2.39
3.6%
54.3
54.3
2.94
33.2%
53.9
55.3
2.14
0.0%
54.6
2.58
6.9%
54.0
3.64
70.0%
47.6
2.35
0.0%
2.34
2.9%
2.65
8.3%
1.99
0.0%
2.56
PF Target Brainste m PF Target Brainste m PF Target Brainste
59.8
1.25
1.08
58.5
1.17
1.10
58.4
1.18
1.13
59.9
1.22
53.9
1.09
57.7
1.16
54.0
1.10
58.7
1.20
50.0
1.16
55.3
1.28
53.9
53.4
1.09
57.8
1.18
54.6
54.6
1.10
58.4
1.18
54.5
55.3
1.12
59.5
1.20
54.7
53.8
1.08
57.3
1.15
SC
1.13
M AN U
8
12.4%
TE D
7
2.63
11.1%
53.4
53.4
1.10
58.0
1.20
71.8%
47.1
49.5
1.16
54.8
1.28
0.0%
53.0
53.0
1.10
57.6
1.19
2.21
0.4%
56.6
56.6
1.10
60.1
1.17
2.68
14.6%
56.0
57.3
1.13
61.2
1.20
1.10
59.0
1.17
3.61 2.55
EP
6
PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC
2.18
AC C
5
RI PT
Non-Symptomatic Treatment Change
0.0% 55.6 55.5 No reported Treatment Change
2.68
11.3%
53.7
53.7
1.10
58.8
1.20
3.73
73.7%
48.1
50.4
1.15
56.2
1.29
2.48
0.0%
55.0
55.0
1.10
59.4
1.19
2.90
39.1%
50.2
51.5
1.13
55.9
1.22
2.54
12.4%
54.9
54.9
1.10
59.5
1.19
3.27
65.1%
47.1
49.1
1.15
53.9
1.26 6
ACCEPTED MANUSCRIPT
15
15
2.32
0.1%
54.0
54.0
1.10
57.8
1.18
3.22
66.5%
41.9
44.5
1.17
48.1
1.26
2.45
2.1%
53.7
53.9
53.7
1.10
57.9
2.85
28.7%
53.2
53.6
54.2
1.12
58.8
2.48
1.4%
53.9
53.9
1.10
58.2
1.19
3.05
40.8%
50.7
52.1
1.13
56.8
1.23
15
RI PT
14
m PF Target Brainste m PF Target Brainste m PF Target Brainste m
AC C
EP
TE D
M AN U
SC
PF: Posterior fossa; STC: symptomatic treatment change; NSTC: non-symptomatic treatment change
7
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Median FU
Christopherson 1968-‐ et al (2014) 2008
Merchant et al 1997-‐ (2009) 2007
Medulloblastoma, PNET
Ependymoma
-‐-‐
24 years (5.6-‐ 44.4)
1/236
4.8 months (to imaging), 6.0 months (for symptoms)
52 months (4-‐163)
7-‐year CI 1.6%
1/153
9 and 12 months
5.3 years (0.4-‐ 10.4)
All toxicity: 2-‐year CI 3.8% Grade 3+ toxicity: 2.-‐ year CI 2.1%
1/313
3 months
2 years
28.8 CSI 54 Boost
236
CSI 23.4-‐ 39.6 PF Boost 0-‐36 IF Boost 55.8-‐ 59.4
8/2362 (3.4%)
All tumors: 5-‐year CI 3.7% Infratentorial tumors: 5-‐ year CI 4.4%
153
59.4 or 54
2/1533 (1.3%)
-‐-‐
SC
M AN U
1996-‐ 2009
3/53 1(5.6%)
1/53
53
TE D
Murphy et al (2012)
Medulloblastoma
RI PT
Appendix Table1: Incidence of radionecrosis after photon and proton RT in pediatric patients Median Crude Cumulative Median time dose incidence of incidence of Deaths from to necrosis # of First Author Year Malignancy (range) patients (Gy or symptomatic symptomatic radionecrosis Gy[RBE]) necrosis necrosis (months) PHOTON RT
313
AC C
Ependymoma, craniopharyngioma, 2007-‐ low-‐grade glioma, Indelicato et al 2013 medulloblastoma/ PNET, rhabdomyosarcoma
EP
PROTON RT
54
11/313 (3.5%)
1 One of these cases is osteoradionecrosis of the auditory canal 2 One of these patients was not symptomatic. 3 This refers to the incidence of brainstem necrosis, not overall RT necrosis.
MacDonald et al
2000-‐ 2011
Ependymoma
70
55.8
0/70
0%
N/A
N/A
This series
2002-‐ 2011
Medulloblastoma
111
CSI 23.4 PF 30.6 IF 54.0
4/111 (4%)
RI PT
ACCEPTED MANUSCRIPT
0/111
-‐-‐
SC
5-‐year CI (4%)
AC C
EP
TE D
M AN U
46 months (1-‐11.7 years) 4.2 years (0.9-‐ 10.9 years)
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Table 2: Nervous system disorders Grade Adverse Event 1 2 3 4 Central nervous Mild Moderate symptoms; limiting Sever symptoms; Life-‐threatening consequences; system necrosis symptoms instrumental ADL limiting self care ADL urgent intervention indicated Definition: A disorder characterized by a necrotic process occurring in the brain and/or spinal cord.
5 Death
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 1:
Figure 1: LET volume histograms for brainstem. No observable difference among the patient populations for LET distributions. If anything, the LET seems to be higher for the patients without TCAs, consistent with lower brainstem doses (Figure 2), implying that dose distributions are more important than LET in these cases.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 2:
Figure 2: Constant RBE (=1.1) based DVHs for brainstem structure.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 3:
Figure 3: Variable RBE (CARABE) based DVHs for brainstem structure. No correlation was observed between elevated LET distributions in brainstem and increased risk of radiation injury.
AC C
EP
TE D
M AN U
SC
Comparing LET distributions between Target and treatment change areas: Appendix Figure 4:
RI PT
ACCEPTED MANUSCRIPT
Figure 4: LET distributions for symptomatic treatment change were somewhat higher on average compared to the target LET
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 5:
Figure 5: LET distributions for asymptomatic treatment change were lower on average compared to the target LET
S0360-3016(15)03343-X
DOI:
10.1016/j.ijrobp.2015.09.015
Reference:
ROB 23169
To appear in:
International Journal of Radiation Oncology • Biology • Physics
Received Date: 4 February 2015 Revised Date:
3 September 2015
Accepted Date: 11 September 2015
Please cite this article as: Giantsoudi D, Sethi RV, Yeap BY, Eaton BR, Ebb DH, Caruso PA, Rapalino O, Chen Y-LE, Adams J, Yock TI, Tarbell NJ, Paganetti H, MacDonald SM, Incidence of CNS Injury for a Cohort of 111 Patients Treated with Proton Therapy for Medulloblastoma; LET and RBE Associations for Areas of Injury, International Journal of Radiation Oncology • Biology • Physics (2015), doi: 10.1016/ j.ijrobp.2015.09.015. 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 Incidence of CNS Injury for a Cohort of 111 Patients Treated with Proton Therapy for Medulloblastoma; LET and RBE Associations for Areas of Injury. Drosoula Giantsoudi*, Roshan V. Sethi*, Beow Y. Yeap, Bree R. Eaton, David H. Ebb, Paul A. Caruso,
RI PT
Otto Rapalino, Yen-Lin E. Chen, Judith Adams, Torunn I. Yock, Nancy J. Tarbell, Harald Paganetti, Shannon M. MacDonald
Department of Radiation Oncology (D.G., R.V.S., B.R.E., Y.E.C., J.A., T.I.Y., N.J.T., H.P., S.M.M.), Department of Medicine (B.Y.Y.), Department of Pediatrics (D.H.E.) and Department of Radiology
SC
(P.A.C., O.R.) at the Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts, 02114 USA
Correspondence:
M AN U
Running Title: Correlations of LET and Brainstem toxicity after proton RT
Shannon MacDonald, M.D., Massachusetts General Hospital, Department of Radiation Oncology
Yawkey 112, 54 Fruit Street, Boston, MA 02114 Tel: (617) 726-5184, Fax: (617) 726-3603
TE D
E-mail: [email protected]
* DG and RVS has contributed equally to this work.
CONFLICT OF INTEREST NOTIFICATION: N.T.’s spouse is on the medical advisory board of ProCure.
EP
All other authors deny any real or potential conflicts of interest. FUNDING STATEMENT: R.V.S. was supported by the Doris Duke Charitable Foundation. Research was
AC C
supported by the National Cancer Institute of the National Institutes of Health under Award Number P01CA021239 and the Federal Share of program income earned by Massachusetts General Hospital on C06 CA059267, Proton Therapy Research and Treatment Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. ACKNOWLEDGMENTS: We would like to thank Jonathan Jackson and Tao Song of the Enterprise Research Infrastructure and Services (ERIS) group at Partners Healthcare for their in-depth support and smooth computing cluster operations, upgrades, and fixes.
1
ACCEPTED MANUSCRIPT
SUMMARY This report investigates the incidence of CNS injury in a large cohort of medulloblastoma patients treated with craniospinal irradiation by proton radiotherapy. An overall incidence of 3.6% was
RI PT
found. No clear correlation was found between the sites of toxicity and elevated RBE due to higher
AC C
EP
TE D
M AN U
SC
linear energy transfer values.
ACCEPTED MANUSCRIPT
ABSTRACT BACKGROUND: CNS injury is a rare complication of radiotherapy for pediatric brain tumors, but its incidence with proton radiotherapy (PRT) is less well defined. Increased linear energy
RI PT
transfer (LET) and relative biological effectiveness (RBE) at the distal end of proton beams may influence this risk. We report incidence of CNS injury in medulloblastoma patients treated with PRT and investigate correlations with LET and RBE values.
SC
METHODS: We reviewed 111 consecutive patients treated with PRT for medulloblastoma between 2002 and 2011 and selected patients with clinical symptoms of CNS injury. MRI
M AN U
findings for all patients were contoured on original planning scans (TCAs-treatment change areas). Dose and LET distributions were calculated for the treated plans using Monte Carlo. RBE values were estimated based on LET-based published models.
RESULTS: At 4.2 years median follow up, the 5-year cumulative incidence of CNS injury was
TE D
3.6% for any grade and 2.7% for grade 3+. Three of four symptomatic patients were treated with a whole posterior fossa boost. Eight of ten defined TCAs had higher LET values than the target but statistically non-significant differences in RBE values (p-value=0.12).
EP
CONCLUSIONS: CNS/brainstem injury incidence for PRT in this series is similar to that reported for photon radiotherapy. The risk of CNS injury was higher for whole posterior fossa
AC C
boost than for involved field. Although no clear correlation with RBE values was found, numbers were small and additional investigation is warranted to better determine the relationship between injury and LET.
Key Words: Medulloblastoma, proton, brainstem injury, CNS injury, linear energy transference
1
ACCEPTED MANUSCRIPT
INTRODUCTION As medulloblastoma outcomes have improved markedly, long-term toxicities in survivors are receiving greater attention. CNS radiation injury, most often seen in the brainstem, is a rare
RI PT
but potentially devastating treatment complication(1). PRT is becoming more widely used for medulloblastoma due to better sparing critical CNS structures and organs anterior to the brainstem and spinal column and its potential to decrease late complications. However, at least a portion
SC
and sometimes the entire brainstem is part of the high-dose target volume for PRT and ideal beam arrangement for sparing the cochlea, neuroendocrine structures and temporal lobes usually
M AN U
requires placing the distal end of the treatment field within the brainstem for an involved field (IF) boost and at or just beyond it for whole posterior fossa (PF) treatment. For PRT, it’s necessary to account for the relative biological effectiveness (RBE) compared to photon treatments. Although an average RBE value of 1.1 is used clinically, it’s
TE D
known that RBE varies within a PRT field, depending dose, linear energy transfer (LET) and radiobiological properties of tissues(2). LET is the energy transferred to absorbing tissue per unit track length of a particle reported in keV/µm and maximizes near the end of proton field range.
EP
LET could impact the incidence of CNS injury from PRT if high LET regions are within the brainstem and RBE-weighted-doses are higher than estimated by current treatment planning
AC C
systems. Clinical experience is maturing and software systems have been developed to accurately calculate the LET distribution for a proton plan within a patient facilitating the investigation of potential correlation of clinical outcomes with LET distributions, to further our knowledge of PRT and potentially improve upon treatment planning for future patients. In this study, we review the incidence and location of CNS and brainstem injury in the largest existing medulloblastoma patient cohort treated with PRT. We calculated the LET
2
ACCEPTED MANUSCRIPT
distribution for patients with and without injury and compared LET values of MRI treatment change areas (TCAs) to those for the whole treated volumes.
RI PT
MATERIALS AND METHODS
Patients and clinical data
SC
Between May 2002 and December 2011, 111 pediatric patients with medulloblastoma were treated with CSI PRT at our institution. Eighty-four of these patients were enrolled on
M AN U
prospective institutional trials. Clinical data obtained from charts review included the patient’s sex, ethnicity, date of birth, pathological diagnosis, histological grade, risk status, surgical procedures, extent of surgery, PRT treatment details, neurological symptoms, date of last follow-
TE D
up, disease and survival status, location and date of first observed symptomatic TCA.
Definition of radiation injury
We defined radiation injury as new or progressive CNS symptoms not attributable to
EP
tumor progression. We included all areas of MRI radiographic changes in symptomatic patients. Consistent with previous publications, injury was retrospectively graded using the National
AC C
Cancer Institute Common Terminology Criteria for Adverse Events (version 4) for central nervous system necrosis(1, 3). (See appendix). Not all cases were considered to represent necrosis, which is why the term “injury” is used preferentially in this manuscript. Areas of MRI radiographic changes in asymptomatic patients included in this study were not reported as radiation injury, since they were not associated with any CNS symptoms.
LET and RBE-weighted-dose distribution 3
ACCEPTED MANUSCRIPT
CSI passively scattered PRT was delivered to all patients in the prone position, followed by either an IF boost or whole PF boost. For patients with reported radiographic changes (TCAs), MRI scans were fused with the original planning CT image set. The locations of radiographic
RI PT
changes were then contoured in the original CT by a neuro-radiologist (P.C.) and a pediatric radiation oncologist (S.M.) based on registration and anatomic landmarks and accounting for growth and change.
SC
Each patient’s anatomical and beam geometry were imported, as treated, into TOPAS (TOol for PArticle Simulation – version b8(4)) Monte Carlo system. Using the clinical dose
M AN U
calculation grid, dose-to-water and corresponding dose-averaged LET (LETd) values, in units of keV/µm/(gr/cm3), were scored in voxel-by-voxel basis, as described by Grassberger et al.(5). To evaluate potential clinical implications associated with the LET distribution, voxel-byvoxel RBE and RBE-weighted-doses were calculated for all patients with TCAs and six
TE D
additional randomly selected patients without reported TCAs. Because RBE estimations are associated with considerable uncertainties and variations among different models, two alternative LET-based RBE models were employed here, described by Freese et al(6) (“Wilkens model”) and
EP
Carabe et al(7) (“Carabe model”), respectively. Both models are based on the standard linearquadratic dose-response formalism and assume linear relationship between LET and RBE for a
AC C
given dose, with the slope depending on the α/β ratio of the tissue. In the Wilkens model, an average target RBE value is considered as an input parameter
along with its mean dose and LET values calculated based on the clinical treatment plan information. The model predicts the relative variations in RBE values, while maintaining the target’s average RBE value. In this work, an average target RBE value of 1.1 was assumed as an input. In contrast, the Carabe model predicts absolute RBE values based on the calculated dose and LET distributions for each structure of reference. Differences in tissue response are 4
ACCEPTED MANUSCRIPT
considered via α/β values in both models. For the structures (i.e. normal brain, healthy CNS tissues) under study an α/β value of 2.1 Gy was used in both models (αx=0.0425 Gy-1 and
RI PT
βx=0.0203 Gy-2), as averaged over the range of values found in literature(8, 9).
Statistics
Time to injury was measured from the starting date of PRT until the date of first observed TCA or
SC
was censored at the last follow-up among patients still alive without injury. The cumulative incidence of radiation injury was estimated with death defined as a competing risk and was
M AN U
compared between dose groups using Gray’s test. LET and RBE value comparisons between the necrotic areas or TCAs and the whole boost target volume, were assessed using the Wilcoxon signed-rank test. Data analysis was performed using SAS 9.3 (SAS Inst Inc, Cary, NC) with p-
RESULTS Patients
TE D
values based on a two-sided hypothesis test.
EP
Table 1 summarizes the demographic, tumor and treatment characteristics of all patients with (n=4) and without (n=107) CNS radiation injury. Also listed are those patients who showed
AC C
radiographic treatment change but developed no symptoms (6 out of the 107). The majority of patients with standard risk disease were treated per or on ACNS0331 and the majority of high-risk patients were treated per or on ACNS0332. A small number of younger patients were treated on or per Headstart II or III protocol prior to radiation. Nearly all patients presented with hydrocephalus. Most had resolution of hydrocephalus with removal of tumor and few required a permanent shunt. None of the 4 patients with CNS injury required a shunt. All patients with symptoms of injury had radiographic findings. No patient underwent biopsy. Clinically, follow 5
ACCEPTED MANUSCRIPT
up was considered sufficient to ensure that these were not cases of tumor recurrence. All patients had areas of gadolinium enhancement and larger areas of T2 hyperintensity.
RI PT
[Table 1]
Radiation injury
Ten patients had post-treatment radiographic changes, of whom four patients experienced
SC
CNS radiation injury at a median time of 9 months (range, 8 to 18 months) from the start of PRT. Table 2 summarizes the clinical details of the four symptomatic patients. Patients 1,3, and 4 were
M AN U
treated with chemotherapy on or per ACNS0331 and patient 2 received a modified Head Start II regimen. This patient received 2 cycles of Methotrexate, Cisplatin, Cytoxin, and Etopside. He developed acute renal failure and chemotherapy was modifed to include oral Etoposide, Vincristine and Carboplatin. Three patients developed brainstem injury symptoms with All three patients received
TE D
corresponding MRI lesions (two grade 3 injury; one grade 4).
treatment. Two recovered fully from their symptoms, but the patient with grade 4 injury remains paraplegic and dependent on a tracheostomy and feeding tube. The fourth patient developed
EP
cervical injury symptoms and a corresponding lesion was found at C1-C2, considered as grade 2 injury. This patient received no intervention other than omission of her final cycle of
AC C
chemotherapy and recovered completely. Two patients developed additional symptomatic radiation injury more than a month after the initial injury. Patient 1 developed osteonecrosis of the right temporal bone (internal auditory canal) 16 months after the first injury. In addition, this patient developed an additional lesion consistent with brainstem necrosis 6 years and 8 months after the first injury. Patient 3 developed additional brainstem injury 27.4 months from the start of RT.
6
ACCEPTED MANUSCRIPT
[Table 2]
The 5-year cumulative incidence of CNS radiation injury was 3.6% for grade 2-4 injury,
RI PT
and 2.7% for with Grade 3+ injury (see Table 3). As all patients with brainstem lesions developed Grade 3+ brainstem injury, the 5-year cumulative incidence of brainstem radiation injury/necrosis is also 2.7%. The risk of all or Grade 3+ only radiation injury was not significantly altered by CSI
SC
dose (>23.4 Gy(RBE) vs. < 23.4 Gy(RBE): 0% vs. 5.3%; p=0.179 and 0% vs. 4.0%; p=0.247) or boost dose (<54.0 Gy(RBE) vs. >54.0 Gy(RBE): 3.8% vs. 0%; p=0.693 and 2.9% vs. 0%;
M AN U
p=0.733). The risk of any radiation injury was higher with the use of a whole PF boost (IF boost vs. whole PF boost; 1.4% vs. 7.7%; p=0.094) but the difference in only Grade 3+ injury was not statistically significant (1.4% vs. 5.1%; p=0.259).
TE D
[Table 3]
Studying the suggested risk factors per Indelicato et al(3), for the three patients who
EP
experienced brainstem injury, no brainstem volume received dose equal or higher than 60 Gy (Table 2). However for three patients (patients 1, 3 and 4; all received PF boost) the median dose
AC C
to the brainstem was higher than 52.4 Gy(3). The same was true for all 6 patients with asymptomatic TCAs, although none of the TCAs in these cases were located within the brainstem.
LET and RBE-weighted-dose calculations Table 4 summarizes the simulated mean LET and RBE-weighted-dose values for all patients under study. The mean LET value for the total of 10 symptomatic TCAs is compared to 7
ACCEPTED MANUSCRIPT
the corresponding value for the whole boost target volume, which was the whole PF for three patients and the involved field for one. Eight out of ten symptomatic TCAs showed higher LET values compared to the whole target (range of differences: 0.02-1.4 keV/µm per unit density).
RI PT
The mean LET value, averaged over all 10 symptomatic TCAs, was 2.7 versus 2.4 keV/µm per unit density in the whole boost target volume. Differences in LET values between TCAs and the corresponding whole boost target volume were statistically significant (p-value<0.05).
SC
The mean LET value, averaged over all 6 asymptomatic TCAs, was lower than the corresponding value for the boost target volumes (2.2 versus 2.5 keV/µm/(gr/cm3), p-value<0.05)
M AN U
but the difference in biological effect between these two LET values would be insignificant (~2%). No statistically significant differences were found on the volume of brainstem with LET more that 3keV/µm/(gr/cm3) between the symptomatic, the asymptomatic and the no-TCA patients. No differences were observed in the LET distributions among the three studied patient
TE D
groups (Appendix figures 1-5).
EP
[Table 4]
For all symptomatic patients the Carabe model predicted higher mean RBE-weighted-dose
AC C
for the boost target and TCAs compared to the constant RBE value of 1.1. Figures 1 and 2 show the results for patients 1 and 2 (tables 2 and 4). Two-dimensional RBE-weighted-dose distributions assuming constant (figures 1a and 2a) and variable RBE (Carabe model - figure 1b and 2b) are presented, along with their difference (figures 1c-d and 2c-d). For patient 1, great variations in LET distributions were noticed within the PF target and among the injured areas, as seen by the distribution of the LET volume histograms (LET-VHs - figure 1e) and the mean values in table 4. The RBE-weighted DVHs for these areas followed a similar pattern with the 8
ACCEPTED MANUSCRIPT
LET-VHs: although for constant RBE their mean doses are close or even lower than the mean target dose, all TCAs, except for the resection cavity TCA, showed increased variable-RBE-
[Figure 1]
RI PT
weighted dose compared to the whole target volume.
SC
Similarly for patient 2 (figure 2), the TCA’s mean RBE-weighted-dose (Carabe model) was found to be 3.1 Gy higher than the corresponding value for the target (figures 1c and d and
M AN U
table 4). The LET-VHs for this case (figure 2e) show considerably increased LET distribution for the TCA (brainstem SRC – figure 2) compared to the IF target (CTV). DVH analysis shows higher variable-RBE-weighted-doses for the brainstem TCA for this case compared to the CTV, while for constant RBE the inverse is observed (figure 2f). Color-wash areas in Figure 2f
TE D
represent the uncertainty due to the range of α/β values reported in the literature for the structures of interest (α/β ranging from 1.5 Gy to 2.8 Gy for normal brain anatomy(5)).
EP
[Figure 2]
AC C
Based on the reported values (table 4 - Carabe model), mean RBE averaged over all 10
symptomatic TCAs was 2% higher compared to the corresponding target mean RBE, but not statistically significant (p=0.12). Furthermore, 2% dose deviations are well within the accuracy of clinical dosimetry and considered clinically insignificant. Similar results were observed with the Wilkens model: under the assumption of average target RBE of 1.1, the mean RBE-weighteddose for the TCAs was not statistically different than when assuming constant RBE.
9
ACCEPTED MANUSCRIPT
DISCUSSION: Our five-year cumulative incidence of overall CNS injury is 3.6%. The cumulative incidence of >Grade 2 injury or brainstem injury is 2.7%. Initial affected sites included the
RI PT
brainstem (n=3) and upper cervical spinal cord (n=1) for this cohort.
The incidence of brainstem radiation injury/necrosis encountered in this series is comparable to photon series, which are summarized in the appendix(1, 10-13). In the largest
SC
retrospective review, Murphy et al followed 236 children with embryonal tumors for a median of 52 months(1). The 5-year cumulative incidence of any radiation necrosis was 3.7% among all
M AN U
patients, and 4.4% among those with infratentorial tumors. The second largest series, by Christopherson et al, followed 53 patients with medulloblastoma for a median of 24 years, with a crude incidence of symptomatic radiation necrosis of 5.6%(10). Although these figures are similar to those encountered in our series, there are several limitations to the comparison. In the
TE D
study by Murphy et al, the primary outcome was imaging changes, though seven of eight affected patients were symptomatic. In addition, Murphy et al administered a higher median IF or PF boost dose compared to our series and a relatively aggressive chemotherapeutic regimen(1).
EP
Similar to Murphy, we found that after the first diagnosis of injury or necrosis, subsequent injury can still develop, even after resolution of the first event.
AC C
There is little available data on CNS injury for PRT(14, 15). Indelicato et al reviewed
children treated with PRT for a variety of central nervous system malignancies(3). Among the 313 patients who received a minimum dose of 50.4 Gy(RBE) to the brainstem, the 2-year cumulative incidence of any brainstem injury was 3.8% and of grade 3+ injury was 2.1%. Rates of injury were higher for PF tumors. MacDonald et al followed 70 patients with ependymoma treated with PRT to a median dose of 55.8 Gy(RBE) to areas at high risk for disease recurrence;
10
ACCEPTED MANUSCRIPT
the majority of patients received 54 Gy(RBE) to the initial CTV(14, 15). At a median follow-up of 46 months, there were no cases of radiation necrosis. It is worth noting that Ruben et al found that sensitizing chemotherapy regimens increased
RI PT
the risk of radiation necrosis by fivefold(16). Spearfico et al reported the highest recorded incidence of symptomatic radiation necrosis in a study of children who received myeloablative chemotherapy (followed by peripheral blood stem cell rescue) in addition to photon RT(13). In
SC
our series all patients with grade 3 brainstem injury/necrosis had PF syndrome and notable sensitivities to chemotherapeutic agents, with one patient tested for DNA sensitivity due to severe
M AN U
chemotherapy injury prior to receiving PRT. While this is only an observation it supports the hypothesis that multiple patient and treatment-related factors contribute to the etiology of brainstem necrosis. The influence of other variables, including dose, gross total resection, perioperative complications and shunting, should also not be discounted. In our series, no significant
TE D
effect was found from the CSI dose or the cumulative involved field dose. We did find an association with boost treatment of the whole PF versus the IF. Due to the limited number of symptomatic patients in our series, it is difficult to draw firm conclusions.
EP
However, this finding supports Murphy’s findings that patients who developed necrosis received >50Gy to a larger percentage of the infratentorial brain (92.1% vs. 72.9%; p=0.03)(1) and
AC C
Indelicato’s finding that median dose to the brainstem of >52.4 Gy is a risk factor for brainstem injury. Some centers employ techniques to avoid placing the end of range in the brainstem by extending the field’s range beyond the brainstem(17). Apart from increasing the brainstem volume receiving full dose, these techniques increase the dose to critical structures beyond the brainstem (cochlea, neuroendocrine structures and temporal lobes). Caution is advised in using them as our findings along with Murphy’s and Indelicato’s suggest they may increase the risk of injury. 11
ACCEPTED MANUSCRIPT
Although our statistical sample of symptomatic TCAs was relatively small, the differences in mean RBE values and RBE-weighted-doses, among the three patient groups that were analyzed, were not statistically or clinically significant. Considering the uncertainties of the RBE
RI PT
models, due to reported range of α/β ratios for the structures of interest, significant variability in potential RBE-weighted-dose distributions is possible (figure 1f). Although variable-RBEweighted DVHs show higher organ doses than those assuming constant RBE value, overlapping
SC
areas of the DVH bands (blue and red) in figure 1f suggests significant uncertainty in the relative RBE-weighted dose difference between TCAs and whole target volumes.
M AN U
In summary, no current evidence exists supporting higher incidence of CNS radiation injury in medulloblastoma patients treated with PRT compared to photon RT. We found higher rate of brainstem injury/necrosis for medullobastoma patients compared to ependymoma patients treated with protons despite comparable doses to similar region. However, we should recognize
TE D
that radiation doses for both these patient cohorts were relatively conservative, rarely exceeding a prescription dose of 54 Gy(RBE) to the CTV and making every effort to avoid brainstem doses exceeding 54 Gy(RBE). Although LET levels were higher in symptomatic TCAs, no clear
EP
correlation was found between the sites of injury and elevated RBE due to higher LET values. Brainstem LET distributions were similar in all patients under study. Furthermore, the greater
AC C
likelihood of brainstem injury in treatment of PF versus IF indicates that the treated volume may play a more significant role for these cases. PRT is a promising treatment for children, with known physical properties that spare
healthy tissues from high and low radiation doses. When administering conservative doses, brainstem injury rates were comparable with the photon experience.
Future directions may
include testing for additional patient related risk factors for radiation injury, such as DNA
12
ACCEPTED MANUSCRIPT
sensitivity or other biomarkers. Further study of clinical outcomes to better explore biological
RI PT
endpoints is still needed.
REFERENCES
SC
1. Murphy ES, Merchant TE, Wu S, et al. Necrosis after craniospinal irradiation: results from a prospective series of children with central nervous system embryonal tumors. Int. J. Radiat. Oncol. Biol. Phys. 2012;83:e655–60.
M AN U
2. Paganetti H, Niemierko A, Ancukiewicz M, et al. Relative biological effectiveness (RBE) values for proton beam therapy. International Journal of Radiation Oncology*Biology*Physics. 2002;53:407–421. 3. Indelicato DJ, Flampouri S, Rotondo RL, et al. Incidence and dosimetric parameters of pediatric brainstem toxicity following proton therapy. Acta Oncol. 2014:1–7. 4. Perl J, Shin J, Schümann J, et al. TOPAS: An innovative proton Monte Carlo platform for research and clinical applications. Med. Phys. 2012;39:6818–6837.
TE D
5. Grassberger C, Trofimov A, Lomax A, et al. Variations in Linear Energy Transfer Within Clinical Proton Therapy Fields and the Potential for Biological Treatment Planning. International Journal of Radiation Oncology*Biology*Physics. 2011;80:1559–1566.
EP
6. Frese MC, Wilkens JJ, Huber PE, et al. Application of constant vs. variable relative biological effectiveness in treatment planning of intensity-modulated proton therapy. Int. J. Radiat. Oncol. Biol. Phys. 2011;79:80–88. 7. Carabe A, Moteabbed M, Depauw N, et al. Range uncertainty in proton therapy due to variable biological effectiveness. Phys. Med. Biol. 2012;57:1159–1172.
AC C
8. Meeks SL, Buatti JM, Foote KD, et al. Calculation of cranial nerve complication probability for acoustic neuroma radiosurgery. International Journal of Radiation Oncology*Biology*Physics. 2000;47:597–602. 9. Hornsey S, Morris CC, Myers R, et al. Relative biological effectiveness for damage to the central nervous system by neutrons. International Journal of Radiation Oncology*Biology*Physics. 1981;7:185–189. 10. Christopherson KM, Rotondo RL, Bradley JA, et al. Late toxicity following craniospinal radiation for early-stage medulloblastoma. Acta Oncol. 2014;53:471–480. 11. Fouladi M, Chintagumpala M, Laningham FH, et al. White matter lesions detected by magnetic resonance imaging after radiotherapy and high-dose chemotherapy in children with medulloblastoma or primitive neuroectodermal tumor. J. Clin. Oncol. 2004;22:4551–4560. 13
ACCEPTED MANUSCRIPT
12. Merchant TE, Li C, Xiong X, et al. Conformal radiotherapy after surgery for paediatric ependymoma: a prospective study. The Lancet Oncology. 2009;10:258–266. 13. Spreafico F, Gandola L, Marchianò A. Brain Magnetic Resonance Imaging After High-Dose Chemotherapy and Radiotherapy for Childhood Brain Tumors. International Journal of …. 2008.
RI PT
14. MacDonald SM, Safai S, Trofimov A, et al. Proton radiotherapy for childhood ependymoma: initial clinical outcomes and dose comparisons. Int. J. Radiat. Oncol. Biol. Phys. 2008;71:979– 986. 15. MacDonald SM, Sethi R, Lavally B, et al. Proton radiotherapy for pediatric central nervous system ependymoma: clinical outcomes for 70 patients. Neuro-oncology. 2013;15:1552–1559.
SC
16. Ruben JD, Dally M, Bailey M, et al. Cerebral radiation necrosis: incidence, outcomes, and risk factors with emphasis on radiation parameters and chemotherapy. Int. J. Radiat. Oncol. Biol. Phys. 2006;65:499–508.
AC C
EP
TE D
M AN U
17. Buchsbaum JC, McDonald MW, Johnstone PA, et al. Range modulation in proton therapy planning: a simple method for mitigating effects of increased relative biological effectiveness at the end-of-range of clinical proton beams. Radiat Oncol. 2014;9:2.
14
ACCEPTED MANUSCRIPT
FIGURE CAPTIONS Figure 1: (Patient 1) RBE-weighted-dose for constant(a) and variable RBE(b), axial(c) and
RI PT
sagittal(d) views of their difference, LET-VHs(e) and RBE-weighted DVHs(f). PF: posterior fossa; TCA: treatment change area; RBEw: RBE-weighted; IAC: Inner Acoustic
SC
Canal;
M AN U
Figure 2: (Patient 2) RBE-weighted-dose for constant(a) and variable RBE(b), axial(c) and sagittal(d) views of their difference, LET-VHs(e) and RBE-weighted DVHs(f) for constant (thick dashed lines) and variable RBE (Carabe model) for a range of a/b ratios (thin solid lines: α/β=2.8 Gy; thick solid lines: α/β=2.1 Gy; thin dashed lines: α/β=1.5 Gy).
AC C
EP
TE D
CTV: clinical target volume (boost); TCA: treatment change area; RBEw: RBE-weighted
15
ACCEPTED MANUSCRIPT
10.5 years 6 years - 22 years 0 (0%) 4 (100%)
66 (59%) 45 (41%)
2 (50%) 2 (50%)
7 years 32 months - 22 years 2 (2%) 105 (98%)
2 (33%) 4 (67%)
64 (60%) 43 (40%)
4 (100%) 0 (0%) 0 (0%) 0 (0%)
6 (100%) 0 (0%) 0 (0%) 0 (0%)
92 (86%) 2 (2%) 6 (6%) 7 (6%)
4 (100%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
5 (83%) 1 (17%) 0 (0%) 0 (0%) 0 (0%)
77 (72%) 18 (17%) 10 (9%) 1 (1%) 1 (1%)
76 (68%) 35 (32%)
4 (100%) 0 (0%)
5 (83%) 1 (17%)
72 (67%) 35 (33%)
5 (4%) 1 (1%) 71 (64%) 3 (3%) 2 (2%) 1 (1%) 28 (25%)
0 (0%) 0 (0%) 4 (100%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
0 (0%) 1 (17%) 3 (50%) 0 (0%) 0 (0%) 0 (0%) 2 (33%)
5 (5%) 1 (1%) 67 (63%) 3 (3%) 2 (2%) 1 (1%) 28 (26%)
0 (0%) 0 (0%) 4 (100%) 0 (0%) 0 (0%)
0 (0%) 0 (0%) 5 (83%) 1 (17%) 0 (0%)
3 (3%) 1 (1%) 99 (93%) 3 (3%) 1 (1%)
96 (86%) 2 (2%) 6 (5%) 7 (6%)
TE D
81 (73%) 18 (16%) 10 (9%) 1 (1%) 1 (1%)
AC C
EP
Anaplastic+Desmoplastic Nodular Risk Classification Standard High CSI PRT Dose (Gy[RBE]) 18 22.5 23.4 27 30.6 34.2 36 Boost Field PRT Dose (Gy[RBE]) 50.4 52.2 54.0 55.8 59.4
Patients without injury (including patients with asymptomatic treatment change)(n=107)
8 years 27 months – 14 years 1 (17%) 5 (83%)
3 (3%) 1 (1%) 103 (93%) 3 (3%) 1 (1%)
SC
Gender Male Female Race Caucasian African-American Asian Other/Unknown Histology Classic Anaplastic Desmoplastic
7 years 32 months - 22 years 2 (2%) 109 (98%)
Patients with asymptomatic radiographic treatment change (n=6)
M AN U
Age at Treatment Median Range < 3 years > 3 years
Patients with injury who developed symptoms (n=4)
RI PT
Table 1: Patient characteristics All patients (n=111)
1
ACCEPTED MANUSCRIPT
69 (62%) 42 (38%) 23.4 (5.4 – 36.0)
1 (25%) 3 (75%) (30.6 – 30.6)
3 (50%) 3 (50%) (18.0 – 31.5)
67 (63%) 40 (37%) 19.8 (5.4 – 36.0)
AC C
EP
TE D
M AN U
SC
RI PT
Boost (Gy[RBE]) Involved Field only Whole Posterior Fossa Median (range)
2
ACCEPTED MANUSCRIPT
8.2/ F
SR
23.4
WPF (30.6)
55.00
56.17
7.9
Leg weakness, ataxia
23.4
3
23.0/M
SR
23.4
WPF (30.6)
5
13.4/F
SR
23.4
WPF (30.6)
83
Dysarthria, ataxia
Brainstem
Steroids, Avastin
Brainstem, upper cervical cord
3
HBOT
AliveNED,NRS
Brainstem
4
Steroids, Avastin
AliveAWD,RS
55.00
9.0
54.55
55.85
17.8
Left hemiparesis
54.56
27.4 56.02
AliveNED,RS
None
50.17
AC C
4
Steroids, HBOT
24
M AN U
SR
TE D
6.4/M
3
Right-sided hearing loss*
Right-sided weakness
EP
2
IF (30.6)
Brainstem, cerebellum
Patient Current Status
Right temporal bone (internal auditory canal)
SC
1
Symptoms
Location of Treatment MRI Grade and changes intervention
RI PT
Table 2: Clinical characteristics of patients with injury Whole Patient PF / Median Maximum Time to Patient age Risk CSI Dose IF(D50) Dose to injury ID (y)/ Group [Gy(RBE)] only Dose to Brainstem imaging Sex (Dose Brainstem [Gy(RBE)] (mo) in Gy)
8.9
Weakness
Brainstem, cerebellum Upper cervical cord
Steroids 2
None
AliveNED,NRS
SR, standard risk; HBOT, hyperbaric oxygen therapy; IF, involved-field only boost; WPF, whole posterior fossa boost. Note: patient age is the age at which the patient underwent radiation; NED: No evidence of disease; AWD: Alive with disease; NRS: No residual symptoms; RS: Residual symptoms of toxicity. *Hearing loss developed before imaging evidence of osteoradionecrosis and may be due to Cisplatin and/or cochlear dose from RT
3
ACCEPTED MANUSCRIPT
PF: posterior fossa
AC C
EP
TE D
M AN U
10
SC
RI PT
Table 3: Crude and cumulative incidence of symptomatic necrosis Grades 2-4 Grade 3+ only Crude 5-year p-value Crude 5-year p-value rate incidence rate incidence All patients 4% 4% 3% 3% CSI Dose (Gy[RBE]) 0.179 x >23.4 0% 0% 0% 0% <23.4 4% 5% 3% 4% Whole PF Boost 0.612 0.862 Involved Field only 2% 3% 2% 3% Whole PF Boost 2% 5% 1% 2% Involved Field (Gy[RBE]) 0.693 0.733 < 54.0 4% 4% 3% 3% > 54.0 0% 0% 0% 0%
4
ACCEPTED MANUSCRIPT Table 4: LET, RBE and RBE-weighted (DRBE) dose calculations results
3
4
AC C
2
EP
TE D
M AN U
SC
1
Mean Dose (CARAB E) a/b = 2.1 [Gy(RB E)]
RI PT
Patie nt
Mean Dose Mean (WILKE Mean Mean LET Dose NS) V(LET> RBE Structur [keV/µm/(u (RBE = a=0.042 3) (WILKE e nit 1.1) 5, [%] NS) density)] [Gy(RB b=0.020 E)] 3 [Gy(RBE )] Symptomatic Treatment Change (Injury) PF 2.36 5.8% 51.0 51.0 1.10 Target Brainste 2.77 20.2% 50.5 51.4 1.12 m Cerebell 2.40 0.1% 51.2 51.3 1.10 um STC Inner Acoustic 3.26 77.3% 50.2 52.4 1.15 Canal STC Brainste 3.01 47.4% 50.6 52.2 1.13 m STC 1 Brainste 2.38 0.0% 51.1 51.2 1.10 m STC 2 Resectio n Cavity 2.03 0.0% 51.3 50.5 1.08 STC Brainste 2.51 0.0% 51.0 51.4 1.11 m STC 3 PF 2.61 14.8% 54.1 54.1 1.10 Target Brainste 3.91 80.6% 45.1 48.0 1.17 m Brainste 4.01 96.5% 52.5 55.7 1.17 m STC PF 2.34 3.1% 54.8 54.8 1.10 Target Brainste 3.17 43.8% 52.7 54.9 1.14 m Brainste 2.75 1.9% 54.7 55.8 1.12 m STC Cerebell 2.24 0.0% 55.0 54.8 1.09 um STC PF 2.24 0.9% 54.4 54.4 1.10 Target
Mean RBE (CARAB E)
54.7
1.18
55.5
1.21
55.0
1.18
56.8
1.25
56.4
1.23
54.9
1.18
53.9
1.15
55.2
1.19
59.0
1.20
53.5
1.31
62.1
1.30
58.7
1.18
59.3
1.24
60.1
1.21
58.6
1.17
57.9
1.17 5
ACCEPTED MANUSCRIPT Brainste m Cervical Spine STC
2.47
2.1%
54.3
54.9
1.11
58.6
1.19
2.80
0.0%
38.5
40.5
1.16
43.1
1.23
1.10
58.9
1.20
9
10
11
12
13
54.1
54.1
3.32
51.4%
52.6
54.2
2.20
0.0%
55.1
54.1
2.39
3.6%
54.3
54.3
2.94
33.2%
53.9
55.3
2.14
0.0%
54.6
2.58
6.9%
54.0
3.64
70.0%
47.6
2.35
0.0%
2.34
2.9%
2.65
8.3%
1.99
0.0%
2.56
PF Target Brainste m PF Target Brainste m PF Target Brainste
59.8
1.25
1.08
58.5
1.17
1.10
58.4
1.18
1.13
59.9
1.22
53.9
1.09
57.7
1.16
54.0
1.10
58.7
1.20
50.0
1.16
55.3
1.28
53.9
53.4
1.09
57.8
1.18
54.6
54.6
1.10
58.4
1.18
54.5
55.3
1.12
59.5
1.20
54.7
53.8
1.08
57.3
1.15
SC
1.13
M AN U
8
12.4%
TE D
7
2.63
11.1%
53.4
53.4
1.10
58.0
1.20
71.8%
47.1
49.5
1.16
54.8
1.28
0.0%
53.0
53.0
1.10
57.6
1.19
2.21
0.4%
56.6
56.6
1.10
60.1
1.17
2.68
14.6%
56.0
57.3
1.13
61.2
1.20
1.10
59.0
1.17
3.61 2.55
EP
6
PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC PF Target Brainste m NSTC
2.18
AC C
5
RI PT
Non-Symptomatic Treatment Change
0.0% 55.6 55.5 No reported Treatment Change
2.68
11.3%
53.7
53.7
1.10
58.8
1.20
3.73
73.7%
48.1
50.4
1.15
56.2
1.29
2.48
0.0%
55.0
55.0
1.10
59.4
1.19
2.90
39.1%
50.2
51.5
1.13
55.9
1.22
2.54
12.4%
54.9
54.9
1.10
59.5
1.19
3.27
65.1%
47.1
49.1
1.15
53.9
1.26 6
ACCEPTED MANUSCRIPT
15
15
2.32
0.1%
54.0
54.0
1.10
57.8
1.18
3.22
66.5%
41.9
44.5
1.17
48.1
1.26
2.45
2.1%
53.7
53.9
53.7
1.10
57.9
2.85
28.7%
53.2
53.6
54.2
1.12
58.8
2.48
1.4%
53.9
53.9
1.10
58.2
1.19
3.05
40.8%
50.7
52.1
1.13
56.8
1.23
15
RI PT
14
m PF Target Brainste m PF Target Brainste m PF Target Brainste m
AC C
EP
TE D
M AN U
SC
PF: Posterior fossa; STC: symptomatic treatment change; NSTC: non-symptomatic treatment change
7
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Median FU
Christopherson 1968-‐ et al (2014) 2008
Merchant et al 1997-‐ (2009) 2007
Medulloblastoma, PNET
Ependymoma
-‐-‐
24 years (5.6-‐ 44.4)
1/236
4.8 months (to imaging), 6.0 months (for symptoms)
52 months (4-‐163)
7-‐year CI 1.6%
1/153
9 and 12 months
5.3 years (0.4-‐ 10.4)
All toxicity: 2-‐year CI 3.8% Grade 3+ toxicity: 2.-‐ year CI 2.1%
1/313
3 months
2 years
28.8 CSI 54 Boost
236
CSI 23.4-‐ 39.6 PF Boost 0-‐36 IF Boost 55.8-‐ 59.4
8/2362 (3.4%)
All tumors: 5-‐year CI 3.7% Infratentorial tumors: 5-‐ year CI 4.4%
153
59.4 or 54
2/1533 (1.3%)
-‐-‐
SC
M AN U
1996-‐ 2009
3/53 1(5.6%)
1/53
53
TE D
Murphy et al (2012)
Medulloblastoma
RI PT
Appendix Table1: Incidence of radionecrosis after photon and proton RT in pediatric patients Median Crude Cumulative Median time dose incidence of incidence of Deaths from to necrosis # of First Author Year Malignancy (range) patients (Gy or symptomatic symptomatic radionecrosis Gy[RBE]) necrosis necrosis (months) PHOTON RT
313
AC C
Ependymoma, craniopharyngioma, 2007-‐ low-‐grade glioma, Indelicato et al 2013 medulloblastoma/ PNET, rhabdomyosarcoma
EP
PROTON RT
54
11/313 (3.5%)
1 One of these cases is osteoradionecrosis of the auditory canal 2 One of these patients was not symptomatic. 3 This refers to the incidence of brainstem necrosis, not overall RT necrosis.
MacDonald et al
2000-‐ 2011
Ependymoma
70
55.8
0/70
0%
N/A
N/A
This series
2002-‐ 2011
Medulloblastoma
111
CSI 23.4 PF 30.6 IF 54.0
4/111 (4%)
RI PT
ACCEPTED MANUSCRIPT
0/111
-‐-‐
SC
5-‐year CI (4%)
AC C
EP
TE D
M AN U
46 months (1-‐11.7 years) 4.2 years (0.9-‐ 10.9 years)
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Table 2: Nervous system disorders Grade Adverse Event 1 2 3 4 Central nervous Mild Moderate symptoms; limiting Sever symptoms; Life-‐threatening consequences; system necrosis symptoms instrumental ADL limiting self care ADL urgent intervention indicated Definition: A disorder characterized by a necrotic process occurring in the brain and/or spinal cord.
5 Death
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 1:
Figure 1: LET volume histograms for brainstem. No observable difference among the patient populations for LET distributions. If anything, the LET seems to be higher for the patients without TCAs, consistent with lower brainstem doses (Figure 2), implying that dose distributions are more important than LET in these cases.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 2:
Figure 2: Constant RBE (=1.1) based DVHs for brainstem structure.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 3:
Figure 3: Variable RBE (CARABE) based DVHs for brainstem structure. No correlation was observed between elevated LET distributions in brainstem and increased risk of radiation injury.
AC C
EP
TE D
M AN U
SC
Comparing LET distributions between Target and treatment change areas: Appendix Figure 4:
RI PT
ACCEPTED MANUSCRIPT
Figure 4: LET distributions for symptomatic treatment change were somewhat higher on average compared to the target LET
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
Appendix Figure 5:
Figure 5: LET distributions for asymptomatic treatment change were lower on average compared to the target LET