Locally advanced rectal adenocarcinoma is effectively treated with chemoradiation and surgery; however, 10 to 25% of patients locally recur within or near a previously irradiated field. Proton radiation therapy (PRT) is ideally suited to the problem of reirradiation for locally recurrent rectal cancer (LRRC).
Seven patients with LRRC in or near prior radiation fields were enrolled on this prospective study from March 2010 to February 2011. All patients underwent positron emission tomography (PET)/computed tomography (CT) simulation and were stratified by low volume (clinical target volume<250 cm3, n=4) or high volume (>250 cm3, n=3). Primary endpoints were feasibility and acute toxicity (within 90 days from PRT initiation). Dosimetry was compared using the Wilcoxon signed-rank test. Tumor response was defined according to PERCIST criteria.
Median follow-up was 14 months (4.9–22.6). Median dose of prior RT was 5040 cGy. Mean PRT dose was 6120 cGy (RBE) (range, 4500–6480 cGy). The total dose sum of prior radiation therapy treatment and PRT was 109.8 Gy (RBE) (range, 95.4–151.2). One patient had surgery prior to and 1 after PRT. Six patients received concurrent 5-fluorouracil-based chemotherapy. Bowel volume receiving 10 and 20 Gy, and the dose to 200 and 150 cm3 of bowel were significantly reduced. There were 3 acute grade 3 and 3 late grade 4 toxicities. Four patients were alive at time of follow-up. Six had a metabolic complete response, of whom 2 subsequently locally recurred. One had initial progressive disease. Of 6 symptomatic patients, 3 had complete pain resolution and 3 partial.
This preliminary report indicates that PRT for LRRC provides dosimetric improvements over intensity-modulated radiation therapy, particularly to the bowel. The modest rates of acute and late toxicity as well as the preliminary outcomes of high tumor and symptom control rates are promising.
Locally advanced rectal adenocarcinoma is effectively treated with chemoradiation and surgery with a 5-year overall survival of around 75% . Approximately 10 to 25% of these patients will experience a local-regional recurrence that is within or near a previously irradiated field. These cases present significant challenges when considering curative treatment. Ideally, recurrences undergo resection; however, particularly in the pelvis, surgery is often infeasible due to late effects of prior radiation therapy and/or surgery and anatomic considerations. If surgery is not feasible or gross residual tumor remains after resection, the remaining options are palliative chemotherapy and/or reirradiation. Reirradiation carries significant risk due to the cumulative effects on organs at risk (OARs). The maximum lifetime dose to OARs at which adverse effects outweigh the local control benefit is unknown. Despite these risks, reirradiation is often considered in the management of local-regional recurrent rectal cancer (LRRC).
Small volumes treated with highly conformal radiation may mitigate the risk for toxicity. Highly conformal radiation techniques with photons, such as intensity-modulated radiation therapy (IMRT) , brachytherapy , intraoperative radiation therapy , and stereotactic body radiosurgery [5, 6] have been employed for retreatment in order to treat a limited tumor volume while sparing normal tissues.
Proton radiation therapy (PRT) is ideally suited to the problem of reirradiation for rectal cancer as it delivers highly conformal radiation to a relatively large volume compared to brachytherapy or SBRT. The physics of the proton beam with virtually no exit dose opens new opportunities beyond photon therapy in the reirradiation setting as the distal penumbra of the proton beam is smaller than the lateral penumbra, eliminating dose to tissues beyond the target structure. These tissues are at the highest risk for injury in the setting of reirradiation where they have already received doses close to their maximum lifetime dose. Therefore, we hypothesized that PRT permits the safe and effective treatment of rectal cancer recurrences as it offers dosimetric advantages compared to photon radiation in terms of integral dose and high-dose depositions in OARs abutting the target.
Our center launched a prospective trial for the reirradiation of recurrent malignancies using PRT with the primary endpoints of feasibility and acute toxicity. For each proton plan generated, a physician-approved IMRT plan was also created prior to treatment start. We present a dosimetric comparison of PRT versus IMRT in LRRC and the preliminary toxicities and outcomes of these patients.
Patients and Methods
This prospective study is currently ongoing in accordance with our institution's institutional review board. All patients were adults with solid tumor recurrence in or near prior radiation fields treated at least 3 months prior. Patients were required to have a Karnofsky Performance Status of ≥ 60 and a life expectancy of greater than or equal to 3 months. The trial was designed to enroll 24 subjects in each of five treatment sites (pelvis, abdomen, thorax, extremities, head and neck), half in the low-volume (clinical target volume [CTV] <250 cc) and half in the high-volume (CTV ≥250 cc) cohorts. Patients in this report were enrolled into the pelvic cohort, specifically with a diagnosis of LRRC. All patients had a histologic diagnosis of adenocarcinoma at original diagnosis. The diagnosis of recurrence was based on growth on computed tomography (CT) and/or positron emission tomography (PET)/CT and was confirmed by biopsy in all but one patient. LRCC patients were stratified as either low-volume (n=4) or high-volume (n=3). Primary endpoints were feasibility and acute toxicity, and secondary endpoints were late complications, local control, overall and disease-specific survival, and comparison of dose volume histograms (DVHs) of the PRT plan and the photon plan generated for backup comparison purposes.
Definition of Target Volume
The CTV was defined as the recurrent tumor or tumor bed plus a margin for microscopic disease based on clinical judgment by the treating physician. All patients underwent PET/CT at the time of simulation; all gross disease identified on CT and all fluorodeoxyglucose (FDG)-avid activity identified by visualization was included. A PTV was defined to account for lateral uncertainties (0.5 to 1 cm) per the International Commission on Radiation Units and Measurements (ICRU) Reports 63 and 78 and employed for both modalities.
IMRT and double-scatter proton plans were generated using the Eclipse Treatment Planning System, Version 8.0 (anisotropic analytical algorithm (AAA), scattering models; Varian Medical Systems, Palo Alto, CA). IMRT plans used seven co-planar 6-MV photon fields with a dynamic multileaf collimator (DMLC). IMRT normal-tissue constraints included the small and large bowel, bladder, rectum, and femurs.
For proton double-scattering planning, in addition to the lateral uncertainties included in the PTV, adjustments were made within the beam-design algorithm to account for range uncertainties along the beam direction . The beam range used was equal to the water equivalent path length (WEPL) to the distal margin of the CTV plus 3.5% of the WEPL due to CT uncertainty as well as 3-mm extra range uncertainty due to beam energy, beamline component thickness, and range compensator design. The modulation width was increased accordingly.
For proton plans, an MLC was designed to shape each field laterally and the margin was calculated to include both the position uncertainties and the lateral beam penumbra. A compensator was used to shape the distal portion of the beam to the CTV. The compensator smearing radius was calculated to account for variations in the alignment. Additional adjustments were made in the design of range compensators to improve the dose conformity. This was necessary since the thickness of the range compensator generated was based purely on the WEPL from the body surface to the distal surface of the CTV and did not consider the effect of inhomogeneous scattering on the dose distribution. The dose distribution did not conform to the distal surface of the target volume and manually modified range compensators to improve the dose conformity to the target were warranted. Each proton field was designed to entirely cover the target. All proton plans were designed using a limited number of fields (1 to 3). All dose calculations were performed using a 2.5 × 2.5 × 1.5 mm3 grid where 1. 5mm was the slice thickness of the CT scans.
All IMRT and PRT plans were physician reviewed and approved prior to patient treatment with PRT.
IMRT and proton plans were optimized to deliver the selected dose in 1.8-Gy (RBE) fractions. For proton therapy, we utilized a generic RBE of 1.1, as is standard per the ICRU Report 78 [7, 8]. The dose delivered was with the goal of definitive treatment of gross disease, and was a minimum of 4500 cGy (RBE), with the final dose at the discretion of the treating physician.
Dose-volume Histogram Analysis
DVHs were analyzed for all of the OARs. Planning was deemed feasible if the following DVH constraint protocol guidelines were met: 99% of the CTV covered by 98% of the prescription dose, 98% of the PTV covered by 95% of the prescription dose (within 5%), and a maximum cumulative point dose to the cauda equina of 100 Gy (RBE).
Acute toxicity was defined as any adverse event observed within 90 days from the initiation of radiation therapy that is felt to be directly related to PRT. Toxicities were graded by the National Cancer Institute's Common Terminology Criteria for Adverse Events (CTCAE), version 4.0 . Toxicities were designated by the study Principal Investigator as definitely, probably, possibly, unlikely, or not related to PRT. Adverse events unrelated or unlikely to be related were not reported. Tumor response was defined according to PERCIST criteria . Pain was evaluated on a numeric pain-rating scale from 1 to 10 before initiation of PRT, during PRT, and at every follow up after PRT. Complete response of pain was defined as achieving a 0 on the pain scale and partial response was defined as any decrease in pain level over multiple visits.
The Wilcoxon signed-rank test was used for all statistical comparisons, with p values less than 0.05 considered significant. P values less than 0.001 were truncated and noted as p<0.001.
Patient and Tumor Characteristics
Seven patients with LRRC without other sites of disease were treated with PRT with curative intent from August 2010 to May 2011 (Table 1). Two patients had received radiation therapy twice prior to PRT. Surgery was able to be performed prior to reirradiation in one patient and was planned for after radiation therapy in another. The mean prescribed PRT dose was 6120 cGy (RBE) , and the theoretical summed total dose maximum was 109.8 Gy (RBE) (Table 1).
Figure 1 depicts a patient who underwent incomplete resection of a sacral recurrence, with a residual FDG-avid 2.3-cm mass. All treatment volumes were located within the pelvis, 5 directly atop the sacrum and 2 between the sacrum and lateral pelvic sidewall (Figure 2). In one patient, a single posterior-anterior (PA) field was used but in all other cases a 2- or 3-beam arrangement was used including PA, posterior oblique, and/or lateral fields.
Proton Radiotherapy Dosimetry
There were statistically significant reductions in bowel doses (Table 2 and Figure 3). There was a statistically significant decrease in the V10 by 85% and V20 by 55% as well as in the dose to 200 and 150 cm3 of bowel. There was a trend towards a decrease in maximum left femur dose and a statistically significant decrease in right femur maximum dose with PRT. Both IMRT and PRT plans achieved adequate coverage of the CTV with the median volume treated to 98% of the prescribed dose (V98) equal to over 99% in both IMRT and PRT.
All patients had summary plans created incorporating current and all prior treatments. Figure 4 demonstrates the PRT, IMRT, and treated summed doses in a patient with two prior radiation therapy courses.
Acutely, there were 3 grade 3 events (1 abdominal pain and 3 diarrhea) that were possibly, probably, or definitely related to PRT, all of which resolved, and no grade 4 events (Table 3). Two patients developed a small bowel obstruction, both occurring 5 months after the start of PRT. One obstruction was postsurgical, after an attempt was made to remove residual tumor after PRT. This same patient developed an enterovaginal fistula at 10 months, felt to be due to tumor invading the vagina.
At a median follow-up of 19.4 mos (4.9–30.7), 1 patient had a metabolic complete response (CR) and anatomic partial response (PR) and to date is free of disease. One patient had progressive disease (PD) locally and received palliative chemotherapy. The remaining 5 patients all had a metabolic and anatomic PR, of whom 2 developed a local recurrence. Three patients died at the time of follow-up, of whom 2 had local recurrences and 1 had local control but developed brain metastases (Table 4).
Three patients had complete and 3 had partial resolution of their pain, 1 of whom then had progression.
In the prospective setting, PRT for the reirradiation of LRRC reduces the dose, including the bowel dose, to many critical OARs when compared with IMRT. The acute and late toxicity resulting from proton reirradiation is modest. Early clinical outcomes are promising, with 4 out of 7 patients alive at a median follow-up of 19 months. There were 57% with a durable partial or complete radiographic response, and 83% with pain control. We are encouraged by these data, as patients referred to this reirradiation protocol tend to have large, unresectable tumors and were recommended to not undergo photon reirradiation by the referring provider.
PRT is superior to IMRT in regards to many OARs. It is notable that all IMRT plans were developed for clinical backup and they were physician-approved prior to starting PRT. Therefore, the IMRT plans are highly optimized, making this a robust dosimetric comparison. There were reductions in bowel V10-V20, but a trend towards increased V30-V40 with PRT. This is not unexpected, as most of the target volumes were directly adjacent to the bowel structure and, with the double-scattering proton beam technology utilized, the distal penumbra calculated to mitigate range uncertanties may offer less conformality compared to IMRT. In addition to distal penumbra, degeneration may occur due to variable location and air volume in the bowel. Studies have shown that the volume of bowel receiving a low dose is predictive of acute toxicity in the primary radiation of rectal cancer, such as the Baglan-Robertson model [11, 12]. In the setting of reirradiation, there are no studies thus far that predict acute or late toxicity from cumulative dose over multiple courses of radiation therapy. This study is not powered to analyze for the relationship between gastrointestincal complications and bowel dose; however, such an analysis would be useful in a larger cohort. Nonetheless, we hypothesize that a reduction in volume receiving a low-intermediate dose should correspond with a decrease in acute and late toxicity.
Notably, there was no significant difference in skin dose between PRT and IMRT, which is reassuring as one of the concerns with PRT is the potential for increased skin dose potentially leading to treatment delays.
To maximize the therapeutic window, we took the following considerations into account: first, we exploited the conformality of PRT. While brachytherapy, IMRT, and SBRT techniques may also be helpful in reirradiation, PRT is able to treat a large area and gives virtually no exit dose through normal tissues. Second, surgery was possible in two patients, 1 pre-PRT and 1 post-PRT (aborted). Kim et al. found that LRRC patients who underwent surgery had improved overall survival over those who did not . Surgery can also be helpful by placing spacers, such as omental flap transposition (OFT), to decrease the proximity of the target structure to OARs .
Third, hyperfractionation has been used to decrease tumor bulk and decrease late complications due to the favorable biologic effects of smaller fraction sizes [15, 16], as has been shown in head and neck tumors . Lingareddy et al. followed 52 rectal cancer patients of whom 42% received twice-daily radiation therapy and found that hyperfractionated radiation therapy reduced late toxicity . We used once-daily fractionation, leaving the outstanding question of the relative importance of the physical properties of PRT versus the biologic properties of hyperfractionation. Our fourth consideration was the use of sensitizing concurrent chemotherapy, used in 86%, which may have also contributed to the toxicity profile. As a fifth consideration, we kept the target volume at the minimum necessary, as established in head and neck reirradiation . Advanced imaging (PET/CT) was utilized in this study to help distinguish tumor or tumor bed from surrounding inflammation or edema.
As the proton reirradiation protocol continues to enroll patients, we await further toxicity results and tumor control data. The safety of PRT for the pelvic cohort specifically will be formally established in safety analyses. This preliminary report indicates that proton reirradiation with concurrent chemotherapy for LRRC provides dosimetric improvements over IMRT, particularly with respect to the bowel. The modest rates of acute and late toxicity as well as the preliminary outcomes of high tumor and symptom control rates are promising.
ADDITIONAL INFORMATION AND DECLARATIONS
Meeting: This research was presented by Abigail Berman at 53rd Annual Meeting of the American Society of Radiation Oncology, Miami Beach, FL, October 2–6, 2011.
Conflict of Interest: We do not have any conflicts or potential conflicts of interest to disclose at this time. We did not use any copyrighted information or patient photos. The data presented in this manuscript was acquired in accordance with the policies of the Institutional Review Board at the University of Pennsylvania.
Acknowledgment: Krisi Varillo, BA.