Evaluating changes in radiation treatment volumes from post-operative to same-day planning MRI in High-grade gliomas

High-grade gliomas (HGG), including World Health Organization (WHO) grade III anaplastic astrocytoma (AA) and grade IV glioblastoma multiforme (GBM), occur in approximately 5 cases per 100,000 people in the US. HGG are the most frequently occurring gliomas, with GBM comprising 53.7% of all new cases and AA accounting for 6.7%[1].

Maximum safe resection is standard initial treatment for HGG, and the extent of resection is correlated with improved outcomes[2]. Several studies have shown that post-operative radiation therapy (RT) alone or combined with chemotherapy improves survival in HGG[3, 4]. RT was historically delivered via a whole brain approach. However, Shapiro et al. showed equivalence in survival in patients treated with whole brain RT to 43 Gray (Gy) with a cone-down to 60 Gy, resulting in a shift in treatment paradigm[5]. Further retrospective data from M.D. Anderson for HGG supported the efficacy of reduced field RT[6]. This approach is further supported by patterns of failure studies which, using combinations of biopsy, brain computed tomography (CT), brain magnetic resonance imaging (MRI), and autopsy information, have uniformly indicated that the predominant failure pattern for HGG patients is within a 2-cm margin of the tumor volume[7‐9].

Given the use of reduced field, conformal RT, it is becoming increasingly important to accurately delineate the post-operative tumor bed. Recent standards for RT for HGG as denoted in Radiation Therapy Oncology Group (RTOG) 0825 guidelines recommend use of a CT-based simulation merged to a pre or post-operative MRI[10]. Guidelines on radiation target volume delineation remain unclear with some authors advocating for smaller margins than those described in the protocol[11, 12]. In addition, guidelines on optimal timing of planning MRI are limited as well. The recommended starting time for radiation is 3-5 weeks (21–35 days) after surgery[10]. The brain deformation and change in the tumor bed during this time period is poorly understood. We hypothesized that large changes in the tumor bed and T2/FLAIR abnormality would occur between the time of surgery and initiation of RT. Therefore, a delayed MRI at the time of RT planning may more accurately reflect the tumor bed at time of RT, and potentially allow us to spare more normal tissue with no decrease in local tumor control. The purpose of this study was to prospectively evaluate the changes in treatment volumes between immediate post-operative MRI and same-day MRI at time of RT planning in an attempt to spare more normal tissue.

In total, 16 patients with GBM and 8 with AA were included in the analysis. The median age at the initiation of RT was 57 years old (range 22-78). Please refer to Table1 for patient characteristics. Treatment-planning CT scans of the brain were obtained a median of 17 days after craniotomy (range 7-32) and merged with immediate and delayed MRI for planning (see Table2). While our institution generally prefers treatment planning 2 weeks after craniotomy in an effort to initiate treatment during postoperative week 3, this was not always possible due to extenuating patient circumstances.

Long term survival of patients with HGG remains poor due to local recurrence after treatment, with the vast majority occurring within a 2-cm area of the presurgical tumor margin, emphasizing the importance of accurately defined treatment volumes[7‐9]. Guidelines for treatment planning have been evolving in recent RTOG protocols, which are used as guidelines for treatment planning by many institutions within the US and internationally. While the recent RTOG 0825 protocol required CT-based planning for RT with post-operative MRI fusion to ensure accurate target delineation, only three years earlier RTOG 0525 used pretreatment MRIs for treatment planning[10]. While CT often displays the surgical tumor bed, MRI presents the planning physician with further tools to elucidate tumor bed and surrounding edema[13]. Contrast-enhanced T1 MRI best evaluates residual tumor and areas of blood–brain barrier breakdown, while T2/FLAIR enhancement best evaluates tissue edema[14, 15].

Our data shows that the CTV1, including the resection cavity and surrounding edema, changes extensively in both GBM and AA patients in the small time interval between surgery and RT planning, with same-day planning MRI elucidating a more accurate representation of the lesion and decrease in surgical edema. CTV2 also changed significantly for combined AA and GBM, but only for GBM patients on subset analysis. Twenty-three of 24 CTV1 volumes decreased in size within an average of 17 days between post-operative MRI and treatment planning MRI. This decrease in size can potentially spare proximal organs at risk and spares normal brain tissue from being treated to the initial dose of 46 Gy. In contrast, CTV2 increased in size in 16 of 24 patients, which could potentially result in improved tumor bed coverage during the treatment boost in comparison to immediate post-operative MRI. Nearly a third of all patients had over a 25% change in CTV2 with planning MRI performed in 17 days or less after craniotomy, including 2 patients with a greater than 50% change in their volumes.

Our data also shows that treating with RT based on same-day MRI does not result in increased rates of tumor recurrence outside of the treatment volume, as no failures were seen outside of the CTV volumes based on same-day MRI. In fact, only one patient had disease recurrence locally which was outside of CTV based on post-operative imaging only. Therefore, while less volume of tissue is irradiated when planning is based on same-day MRI versus immediate post-operative MRI, it is unlikely that treatment based on same-day MRI results in an increase in tumor-miss. These changes could result in alterations in treatment volumes, resulting in a large impact for future RTOG and other protocols, which may be based on imaging sets that do not accurately define the tumor bed volume.

Similar findings regarding volume changes were presented at the 2010 American Society for Radiation Oncology Annual Meeting, revealing large variations in surgical cavities with both expansion and shrinkage occurring mostly within the first two weeks after craniotomy for HGG patients. This group of investigators also demonstrated a benefit in local control and a decrease in radiation injury in patients with planning MRI delayed for over two weeks[16]. While we cannot determine a benefit in local control as the aggressive nature of these tumors results in frequent local recurrence, it appears that when MRI is delayed over two weeks, normal tissue can be spared with no detriment on local control.

While RTOG 0825 has a clear set of guidelines concerning treatment margins, some practitioners advocate tighter margins[11, 12]. However, when reviewing our data, caution should be employed with this technique as it introduces the potential of missing gross tumor, as our data indicates that GTV increases even in the short time period between post-operative and same-day MRI, timing of MRI may become more important for practitioners of limited-margin treatment to limit the possibility of tumor miss.

While we have shown that the resection cavity undergoes extensive volumetric changes in the short interval between surgical resection and RT planning, several studies have illustrated volumetric changes throughout treatment. Shukla et al. have shown that at week 5 of treatment, 12/15 patients with unifocal disease had a median reduction of 54.85cc in tumor volumes on T2-weighted MRI[17]. As a result, they suggest mid-treatment re-planning MRI, potentially to define boost fields. Tsien et al. prospectively followed 21 patients who were rescanned at week 3 during RT[18]. They found an increase in GTV in 3/12 patients with GBM, with all cases requiring increase in size of treatment fields. A multi-institutional trial examining volume changes midway through treatment of GBMs showed a change in GTV in 80% of cases[19]. Given the increasing impetus to reduce treatment volumes recently, further research in this area may be warranted.

It is possible that the treatment volume undergoes several changes as the resection space contracts, surrounding edema waxes and wanes, and the actual tumor volume changes, and no minimal amount of reasonable reimaging could monitor this change. While the decrease in treatment volume seen on T2/FLAIR imaging in our study was likely from a reduction in peri-tumor edema, the increase in GTV2 was somewhat unexpected. This change may represent evolution of blood products, principally methemoglobin, post-operative changes which have been documented to occur between days 5 and 14 post-operatively, and/or tumor progression[20]. Optimal imaging times remain unknown but target volumes may be least ambiguous after 14 days postoperatively. As a result, we recommend that the MRI used to define the planning volume should be delayed to the time of CT simulation, with the treatment to start as soon as possible to reduce the potential for further changes that might negatively impact the effectiveness of treatment.

The results of this study illustrate the changes that the post-operative tumor bed of high-grade gliomas undergoes prior to RT and the important effect on target delineation during treatment planning. The marked reduction of volume treated to 46 Gy based on same-day versus post-operative MRI may result in better sparing of normal brain tissue without a detriment in local control. Additionally, the increased CTV2 volume may result in better tumor coverage, avoiding potential geographic miss in treatment based on immediate post-operative MRI scan. Given the differences demonstrated in treatment volume at varying MRI time points, further attempts should be made to correlate treatment planning with patterns of failure to maximize local control while minimizing toxicity.

Research was supported in part by the Kimmel Cancer Center’s NCI Cancer Center Support Grant P30 CA56036.