Module 5: Proton Therapy for Gastrointestinal Cancers: Review of the Evidence
The goal of this module is to review the dosimetric and clinical evidence around proton beam therapy (PBT) for esophageal, pancreatic, and anal cancers.
Esophageal Cancer
Radiotherapy plays an integral role for patients with esophageal cancer, both in resectable and unresectable cases. Nearby organs at risk include the spinal cord, heart, and lungs.
Dosimetric Studies
Several studies have analyzed dosimetric differences between IMRT (intensity modulated x-ray therapy) and PBT for esophageal cancer. Each study used unique planning techniques, but in general the results suggest PBT can limit radiation to nearby organs at risk.
Wang and colleagues investigated dosimetric differences between IMRT and PBT for 55 patients. Most cases were distal esophageal tumors, and all received 50.4 Gy. IMRT was planned with a 7-field technique. PBT was delivered by passive scattering using two beams, typically in a left lateral and posteroanterior arrangement.. Four-dimensional CT planning accounted for respiratory motion. The results showed less radiation to the cord, lungs, and heart with PBT. For the cord, PBT reduced the maximum dose by 8.4 Gy. For the lungs, PBT reduced the mean dose by 2.9 Gy and the volume of spinal cord receiving 5 Gy, 10 Gy, and 20 Gy (V5, V10, and V20) by 20%, 10% and 4% (absolute differences), respectively. For the heart, PBT reduced the mean dose by 0.5 Gy and the V10, V20, and V30 by 33%, 13%, and 3% (absolute differences), respectively. The heart V40 was 2.5% higher (absolute difference) with PBT.
Zhang and colleagues evaluated IMRT and PBT plans for 15 patients with distal esophageal tumors treated to 50.4 Gy. IMRT plans used 5 coplanar beams. PBT was delivered by passive scattering using either a two-beam
anteroposterior-posteroanterior (AP/PA) arrangement or a three-beam technique. Four-dimensional CT planning accounted for respiratory motion. The results showed that the three-beam PBT plans achieved the lowest cord dose compared to IMRT (15 Gy absolute reduction in cord maximum). The two-beam PBT plans maximally reduced the lung dose compared to IMRT (5 Gy absolute reduction in mean dose; 36%, 21%, and 6% absolute reduction in V5, V10, and V20, respectively). This lung sparing came at the cost of higher heart dose with two-beam PBT compared to IMRT (6% and 14% absolute increase in V40 and V50, respectively). The three-beam PBT plans best reduced the heart dose compared to IMRT (8% absolute reduction in V40, no difference in V50).
Welsh and colleagues compared IMRT with PBT in 10 patients with distal esophageal tumors. IMRT was planned with a 5-field step-and-shoot technique, using a simultaneous integrated boost to deliver 65.8 Gy to the gross tumor volume (GTV) and 50.4 Gy to the planning target volume (PTV) in 28 fractions. The same doses were delivered in the PBT plans, which used an intensity-modulated proton technique. Three different sets of PBT beam arrangements were explored: (1) anteroposterior-posteroanterior (AP/PA), (2) right posterior and left posterior oblique (RPO and LPO), and (3) AP combined with RPO and LPO. Four-dimensional CT planning accounted for respiratory motion. The results showed that the three-field AP/RPO/LPO PBT plans achieved the lowest cord dose compared to IMRT (12 Gy absolute reduction in cord maximum). AP/PA PBT plans maximally reduced the lung dose compared to IMRT (5 Gy absolute reduction in mean dose; 26%, 13%, and 7% absolute reduction in V5, V10, and V15, respectively). RPO/LPO PBT plans best reduced the heart dose compared to IMRT (9 Gy absolute reduction in mean dose; 50%, 22%, and 8% absolute reduction in V10, V20, and V30, respectively).
Clinical Studies
There are no comparative clinical studies of photons versus protons for esophageal cancer. However, several groups have published single-institution experiences with PBT.
Lin and colleagues reported the MD Anderson experience of 62 patients treated with concurrent chemotherapy and PBT. The cohort was predominantly stage II or III distal adenocarcinomas. The chemoradiotherapy was definitive for about half of patients, and the other half received neoadjuvant treatment prior to surgery. PBT was delivered with a passive scattering technique, and several beam arrangements were used (two-beam anteroposterior-posteroanterior; two-beam posteroanterior plus left lateral oblique; and three-beam left lateral, left posterior oblique and either a right posterior oblique or posteroanterior beam). Four-dimensional CT planning accounted for respiratory motion. The median prescription was 50.4 Gy, and the most common concurrent chemotherapy regimen was 5-FU, oxaliplatin, and taxol. A single patient had to stop treatment due to esophagitis. Grade 3 toxicities included esophagitis (10%), dysphagia (10%), nausea/vomiting (8%), fatigue (8%), anorexia (5%), and radiation dermatitis (3%). There was a single case of grade 3 pneumonitis. Two patients died during treatment, one from cardiac arrest and one from pneumonitis. With a median 20-month follow-up for living patients, the estimated 3-year local-regional control rate was 41%. Of the patients who underwent surgery, a complete pathologic response was seen in 28%.
The University of Tsukuba in Japan has also published clinical results of PBT for esophageal cancer. One series examined only squamous cell carcinoma treated with radiotherapy alone; the relevance of these results to current US practice (where concurrent chemotherapy is typically given) is challenged. A second publication from the Tsukuba group examined 40 patients treated with concurrent chemotherapy and PBT with definitive intent. Tumors were generally in the upper or middle thoracic esophagus and stage I-III. PBT was delivered with a respiratory-gated technique to a total dose of 60 Gy. Anteroposterior-posteroanterior beams were used for the first 40-50 Gy, followed by either the same fields or an anterior and lateral oblique arrangement for the conedown. Chemotherapy consisted on 5-FU and cisplatin. No patient needed to stop radiotherapy due to toxicity. Grade 3 toxicities included esophagitis (22%) and dermatitis (5%). There was a single grade 2 and no grade 3 late pulmonary complications. No treatment-related deaths were observed. With a median 2-year follow-up, 68% of patients achieved local control.
As a rough comparison, we can examine the toxicities reported for two recent phase III trials examining neoadjuvant chemoradiotherapy using photons followed by surgery versus surgery alone. Twenty-eight patients treated to 50.4 Gy with 3D conformal photons and cisplatin/5FU on CALGB 9781 had toxicity data reported. Grade 3 toxicities included esophagitis/dysphagia (27%), pain (16%), other GI (14%), nausea (11%), weight loss (11%) and dysrhythmias (8%). Meanwhile, 171 patients treated to a notably lower dose of 41.4 Gy with 3D conformal photons and carboplatin/paclitaxel on the CROSS trial had toxicity data reported. Grade 3 or higher toxicities included anorexia (5%), esophagitis (1%), esophageal perforation (1%), fatigue (1%), and nausea/vomiting (1%). Pathologic complete response rate was 29% – very similar to the results from the MD Anderson PBT experience.
Open clinical trials
- A phase I dose escalation trial is open at the University of Pennsylvania (clinicaltrials.gov identifier NCT02213497). This study is enrolling patients with adenocarcinoma distal to the carina. Treatment consists of neoadjuvant PBT concurrent with carboplatin/paclitaxel followed by surgery. The main endpoint is the maximally tolerated radiation dose.
- A phase II trial is open at Loma Linda University (clinicaltrials.gov identifier NCT01684904). This study is enrolling patients with squamous cell carcinoma or adenocarcinoma of the mid or distal esophagus. Treatment consists of neoadjuvant PBT concurrent with cisplatin/paclitaxel followed by surgery. The main endpoints are overall survival and toxicity.
- A phase III trial is open at MD Anderson Cancer Center (clinicaltrials.gov identifier NCT01512589). This study is enrolling patients with either squamous cell carcinomas or adenocarcinomas. Patient can have resectable or unresectable disease, and treatment consists of concurrent chemoradiotherapy to 50.4 Gy. Patients are randomized to either IMRT or PBT radiotherapy. The main endpoints are progression-free survival and toxicity.
Pancreatic Cancer
The benefit of radiotherapy for pancreatic cancers is controversial – both in unresectable cases and following surgery. However, radiotherapy still plays a role for many patients in the United States. Nearby organs at risk include bowel, liver, stomach, and kidneys.
Dosimetric Studies
Several studies have analyzed dosimetric differences between IMRT and PBT for pancreatic cancer. Each study used unique planning techniques, but in general the results suggest that PBT can reduce low doses to many organs at risk. However, PBT may incrementally increase high-dose irradiation of the small bowel. Notably, severe bowel toxicity is thought to correlate to high-dose exposure.
Thompson and colleagues compared IMRT and PBT plans for 13 patients with unresectable pancreatic head tumors. Targets were the gross tumor with a 1-inch margin to PTV; elective nodal regions were not included. The total dose was 55 Gy in 25 fractions. IMRT was planned with a 7-field technique. PBT was planned with a three-beam arrangement (right superior posterior oblique, posterior, and posterior superior oblique). Both double-scattering and pencil-beam scanning PBT were evaluated. Plans were generated on static CT images under the assumption that patients would be treated with a breath-hold technique. The results showed that pencil-beam scanning PBT was superior to double-scattering PBT in reducing dose to organs at risk. However, the conclusions were mixed regarding the benefit of PBT relative to IMRT. For the kidneys, PBT reduced mean dose by 1.7 Gy compared to IMRT. PBT also resulted in a 50% relative reduction in mean liver dose, noting that IMRT was able to meet liver constraints for all patients. For the small bowel and stomach, PBT was superior in the low-dose regions (< 30 Gy) but IMRT was incrementally superior in the high-dose regions (> 40 Gy). For example, small bowel V20 was 20% vs 7% for PBT vs IMRT, but the V45 was 2.9% vs 2.4% for PBT vs IMRT.
Nichols and colleagues evaluated IMRT and PBT plans for 8 pancreatic patients in the post-surgery setting. The dose was 45 Gy to a larger post-operative field followed by a conedown to a total 50.4 Gy. IMRT was planned with 10 to 18 fields. PBT employed a passive scattering technique with either a two-beam or three-beam arrangement (one or two posterior oblique fields with a lateral oblique field). The results showed that PBT reduced the low-dose exposure for several organs, but there were mixed consequences for small bowel. There was no significant difference between modalities for the liver or the left kidney. The right kidney was better spared with PBT vs IMRT (mean V18 23% vs 51%). Low dose to the stomach was reduced with PBT versus IMRT (mean V20 2% vs 20%) but not different at higher doses (≥ 45 Gy). For the small bowel, PBT was superior in the low-dose regions but IMRT was incrementally superior in the high-dose regions. For example, small bowel V20 was 15% v 47% for PBT vs IMRT. But the V45 and V50 were 8% vs 6% and 6% vs 3% for PBT vs IMRT, respectively.
Ding and colleagues analyzed dosimetric differences between IMRT, volume-modulated arc therapy (VMAT), and PBT for 11 patients in the post-operative setting. Four-dimensional CT planning accounted for respiratory motion, and the dose was 50.4 Gy. IMRT used a 5-field approach, while VMAT was planned with 2 arcs. Both passively scattered and pencil-beam scanning proton techniques were analyzed, each with a two-beam arrangement (right lateral oblique and posterior oblique). The results showed that pencil-beam scanning PBT was superior to the passively scattered approach. PBT achieved lower kidney V18 and lower mean liver dose compared with IMRT or VMAT, although there were no differences in liver V30. Stomach V20 was 7% vs 26% for pencil-beam PBT vs IMRT. There were no differences among the photon and proton plans for absolute volume of small bowel receiving ≥ 15 Gy.
Clinical Studies
There are no comparative clinical studies of photons versus protons for pancreatic cancer. However, several groups have published single-institution experiences with PBT.
Nichols and colleagues reported the University of Florida experience of 22 patients treated standard fractionation PBT and concurrent capecitabine. Unresectable patients (n = 5) received a median 59.4 Gy, borderline resectable patients (n = 5) received 50.4 Gy, and post-operative patients (n = 12) received a median 54 Gy. Four-dimensional CT planning accounted for respiratory motion. The PBT technique (i.e. passive scatter, pencil-beam scanning) was not explicitly defined. Two- or three-beam arrangements were used, typically with one or two posterior oblique fields plus a lateral oblique beam. No patient needed to stop radiotherapy due to toxicity. With a median 11-month follow-up, there were no acute or long-term grade 3 treatment toxicities. The tumor control outcomes from this study are difficult to interpret given the heterogeneous population.
Terashima and colleagues reported a phase I/II Japanese trial of 50 locally advanced patients treated with hypofractionated, dose-escalated PBT and concurrent gemcitabine. The primary tumor, positive nodes, and regional elective nodes were targeted, and respiratory gating was used. Neither the PBT technique (i.e. passive scatter, pencil-beam scanning) nor the beam arrangements were explicitly defined. The majority (n = 40) of the cohort received 67.5 Gy in 2.7 Gy fractions with a field-in-field technique (1.8 Gy to the whole PTV with additional 0.9 Gy to PTV minus GI tract). For these 40 patients, 13% had to stop treatment due to side effects. Acute grade 3 non-hematologic toxicities included anorexia (8%), nausea (5%), abdominal pain (5%), weight loss (5%), vomiting (3%), and fatigue (3%). Late grade ≥3 gastric ulcers occurred in 10% of the cohort; one patient died from gastric hemorrhage. The one-year local control rate was 80%.
Hong and colleagues reported a Harvard phase I/II trial of 50 resectable patients treated neoadjuvantly with short-course hypofractionated PBT plus capecitabine followed by surgery. The gross tumor as well as regional elective nodes were targeted, and four-dimensional CT planning accounted for respiratory motion. Passively scattered PBT was delivered to 25 Gy in 5 fractions for 35 patients in the trial’s phase II component. A three-beam arrangement was typically used, treating 2 fields each day. None of the 35 patients needed to stop treatment due to toxicity. Grade 3 toxicities were colitis (3%) and chest wall pain (3%). No grade 4 or 5 toxicities were reported. The rate of negative-margin resection was 84%. With a median 38-month follow-up, local control was achieved in 84% of patients.
As a rough comparison, we can examine toxicity on prospective studies treating pancreas cancer with photons. RTOG 9704 treated locally advanced patients to 50.4 Gy with 3D conformal photons along with 5-FU. Acute non-hematologic grade ≥ 3 toxicities occurred in 59% of the cohort17; late grade ≥ 2 non-hematologic effects were seen in 18%. An ECOG trial treated locally advanced patients to 50.4 Gy with 3D conformal photons along with gemcitabine. Acute non-hematologic toxicities were common (41% grade 3, 38% grade 4, and 3% grade 5). The University of Michigan phase I/II trial treated unresectable patients with IMRT to 50-60 Gy with gemcitabine. Eleven cases (22%) developed dose-limiting toxicities, including grade 3 or 4 nausea/vomiting, anorexia, dehydration, duodenal bleeding or perforation. Two patients had possibly treatment-related fatal toxicities.
Open clinical trials
- The University of Florida has two open phase II trials using protons for pancreatic cancer (clinicaltrials.gov identifiers NCT00763516 and NCT01553019). Treatment consists of PBT to 50.4 Gy along with capecitabine for borderline resectable patients, and PBT to 50.4 Gy alone with gemcitabine for post-operative patients. The main endpoints are rates of grade ≥ 3 toxicity.
- A phase II trial is open at Loma Linda University (clinicaltrials.org identifier NCT01683422). This study is enrolling patients with locally advanced disease. Treatment consists of PBT along with gemcitabine, erlotinib, and capecitabine. The main endpoint is the one-year survival rate.
- A phase II trial is open at Massachusetts General Hospital (clinicaltrials.gov identifier NCT01821729). This study is enrolling patients with unresectable disease. Treatment consists of PBT along with FOLFIRINOX and losartan. The main endpoint is feasibility of the regimen.
Anal Cancer
Most patients with localized anal cancer are cured with chemoradiotherapy. However, the treatment carries high rates of side effects. Nearby organs at risk include bowel, genitalia, and pelvic bone marrow.
Dosimetric Studies
Ojerholm and colleagues investigated dosimetric differences between IMRT and PBT for 8 patients with anal cancer. Tumor T category ranged from T1-T4, and half of patients had positive nodes. IMRT was planned with a 7-field sliding-window technique. PBT employed pencil-beam scanning with a two-beam left and right posterior oblique field arrangement. The most common dose was 54 Gy to the primary tumor/positive nodes and 45 Gy to uninvolved pelvic nodes. The results showed that PBT reduced radiation to most organs at risk. Compared to IMRT, PBT reduced small bowel V15, V20, and V25 by an average of 70 mL, 59 mL, and 36 mL, respectively. The external genitalia mean dose was 19 Gy vs 7 Gy for IMRT vs PBT. The external genitalia V20 was 40% vs 14% for IMRT vs PBT. The total pelvic bone marrow was exposed to less radiation with PBT for doses up to 30 Gy; however PBT was inferior in sparing the lumbosacral marrow component due to the posterior oblique beam arrangement.
Clinical Studies
There are currently no published clinical results with modern PBT for anal cancer.
Open clinical trials
- A phase 0 pilot study is open at the Massachusetts General Hospital (clinicaltrials.org identifier NCT01858025). This study is enrolling patients with all stages of anal cancer. Treatment consists of pencil-beam scanning PBT with concurrent chemotherapy (5-FU and mitomycin-C). The main endpoints are feasibility and toxicity.