To review early outcomes for patients enrolled on our institution's protocol (PC01) for patients with unresectable pancreatic cancer, and to test whether the serious adverse event rate could be reduced from 15% (expected) to <5%.
Twelve patients were enrolled, but only 11 patients are reported in this analysis. Pathology on all patients was reviewed at University of Florida Health Medical Center in Jacksonville to confirm the diagnosis of pancreatic adenocarcinoma. Unresectability was defined by radiographic evidence of encasement of the celiac axis and/or superior mesenteric artery or by occlusion of the superior mesenteric vein, portal vein, or both confluences. Patients received proton therapy to a planning target volume dose of 59.4 Gy (relative biological effective) at 1.8 Gy (relative biological effective) per fraction over 7 weeks with concomitant oral capecitabine (1000 mg orally twice-daily, 5 days/week on radiation treatment days only). Only gross disease was targeted.
Median follow-up for all patients was 14 months (range, 5 to 25 months). Median follow-up for surviving patients was 23 months (range, 8 to 25 months). Median age was 68 years (range, 51 to 86 years). One- and 2-year overall survival, progression-free survival, distant metastasis-free survival, and freedom from local progression rates were 61% and 31% (median, 18.4 months), 55% and 14%, 68% and 27%, and 86% and 69%, respectively. No patient experienced a grade 3 or greater toxicity during treatment or follow-up. Grade 2 toxicity was limited to a single patient experiencing grade 2 fatigue. Median weight loss over the course of treatment was 1.7 kg (range, loss of 5.7 to gain of 4.9 kg). Four patients had an adequate radiographic response to radiation therapy to justify surgical exploration.
Proton therapy with concomitant capecitabine as delivered on the PC01 protocol was well-tolerated with no grade 2 or greater gastrointestinal toxicities. The 18.4-month median survival and the 69% freedom from local progression at 2 years are encouraging. The lack of meaningful gastrointestinal toxicity suggests significant opportunities for treatment intensification when proton therapy is used in this setting.
Conventional x-ray-based radiation therapy, with or without concomitant chemotherapy, for patients with unresectable pancreatic cancer is associated with poor survival. The median survival rates for the radiation therapy arms in 5 prospectively randomized series range from 8.3 to 11.1 months [1–5].
Effective radiation therapy delivery for these patients is compromised by the fact that x-ray beams directed at the critical tumor target must often pass through the spinal cord, liver, and kidneys. Additionally, the pancreas is intimately surrounded by highly radiosensitive structures including the stomach, duodenum, and jejunum. As a result, it is difficult to deliver conventionally fractionated radiation therapy doses much in excess of 54 Gy, particularly if radiosensitizing chemotherapy is utilized. Although, historically, radiation therapy combined with chemotherapy has been offered as a standard treatment for patients with unresectable pancreatic cancer, results from a large recent randomized study was unable to identify a survival benefit for chemoradiation therapy compared to treatment with chemotherapy alone .
Various dosimetric studies have suggested that proton therapy, by virtue of its ability to create more conformal radiation therapy dose distributions, may be able to improve the therapeutic index for patients with pancreatic cancer when compared to treatment with x-ray-based therapies [7–10]. A prior publication from our institution including patients with unresectable cancer, marginally resectable cancer, or those treated postoperatively demonstrated low gastrointestinal toxicities for patients treated with proton therapy and concomitant oral capecitabine . Herein we present the first reported outcomes for a prospective clinical trial utilizing conventionally fractionated proton beam radiation therapy with concomitant chemotherapy for patients with unresectable pancreatic cancer.
Patients and Methods
The University of Florida Proton Therapy Institute PC01 protocol received initial institutional review board approval on July 27, 2007. The first patient was enrolled in April 2010. The accrual target was 66 patients to test if the expected serious adverse event rate could be reduced from the expected 15% to < 5%. Unresectability was defined by radiographic evidence of encasement of the celiac axis, the superior mesenteric artery, or both or by occlusion of the superior mesenteric vein, portal vein, or both confluences. Protocol therapy consisted of proton therapy to a planning target volume dose of 59.4 Gy (relative biological effective [RBE]) at 1.8 Gy (RBE) per fraction over 7 weeks with concomitant oral capecitabine (1000 mg orally twice daily, 5 days per week on radiation treatment days only). Only gross disease was targeted.
Technical Aspects of Treatment Delivery
All patients underwent a 4-dimensional computed tomography scan with intravenous and oral contrast for treatment planning. Although the protocol allowed for the use of fiducial markers, in practice, they were not utilized for patient positioning during treatment. An internal gross tumor volume (IGTV) was established for all patients. A clinical target volume was created by expanding the IGTV by between 3 and 10 mm. A planning target volume was established for all patients providing a margin between 5 and 10 mm around each clinical target volume to compensate for the variability of treatment setup and internal motion. In summary, the minimum expansion on the IGTV was 8 mm and the maximal expansion was 20 mm. Weekly verification scans were performed on all patients during treatment to assess target motion and, as necessary, allow for adaptive replanning over the course of proton therapy. Critical normal structures at risk of toxicity were contoured, including the right and left kidneys, liver, small bowel, and spinal cord. Nuclear medicine renal scans were performed on all patients to assess the bilaterality of renal function. Normal-tissue constraints were established to assure that at least two-thirds of the volume of one functioning kidney received a dose of less than 20 Gy (RBE). The hepatic V30 Gy (RBE) was limited to 50% and the V35 Gy (RBE) was limited to 33% of the liver volume. The small intestine volume was established as a “small bowel space.” Individual loops of bowel were not contoured. Protocol guidelines indicated that no more than 10% of the total small intestine or gastric volumes should receive greater than 50 Gy (RBE) and that no more than 15% of the volume should receive greater than 45 Gy (RBE). Duodenal mean dose was suggested to be < 50 Gy (RBE). In practice, a retrospective review indicated that the entire duodenum in most patients receive the entire prescription dose. Maximum spinal cord dose was limited to 45 Gy (RBE) to 0.1 cm3.
Passive double-scattering proton beams were used to deliver the proton therapy treatments, utilizing apertures and range compensators for lateral and distal dose distribution conformity, respectively. While the double-scattering technique is unable to provide dose conformity in the entrance region to the treatment target, the rapid formation of the spread-out Bragg peaks from our proton therapy system (IBA, Louvain la Neuve, Belgium), at a frequency of 10 Hz, renders the dose delivery insensitive to organ motion, particularly with moving targets such as the pancreas. All patients were treated utilizing a 2-field technique consisting of a heavily weighted posterior-anterior field combined with a single right lateral or right lateral oblique field. For most patients, 75% of the dose was delivered through the posterior field while respecting the constraints on the spinal cord dose, which was limited to 45 Gy (RBE) as described above. The remainder of the dose was delivered through the right lateral field. Anterior and left lateral fields were avoided to minimize small bowel and gastric exposure. The choice to use a posterior-anterior field that penetrates the patient's spine before reaching the target maximizes the robustness of the delivered dose distribution due to the minimal daily anatomical variations that may perturb the delivered beam ranges. Range uncertainties for the lateral treatment field, which typically passes through the liver, could be caused by liver motion and the daily variations of small bowel contents. Robustness of the lateral treatment field was carefully evaluated by calculating the verification treatment plans with the bowel contents overridden alternatively with air or water for the planned dose distribution to meet the planning goals when either air or water was present in the small bowels. A typical plan is shown in Figure 1.
Chemotherapy consisted of capecitabine at a fixed dose of 1000 mg orally twice a day on days when radiation therapy was delivered. Patients were asked to maintain a pill diary documenting self-administration of the drug. The diary was reviewed with nursing on a weekly basis. Standard chemotherapy dose modification guidelines were utilized although there were no protocol patients who were unable to receive chemotherapy.
Patients were allowed to receive chemotherapy before chemoradiation as long as the regimens did not cause hematologic toxicity that would prohibit protocol enrollment. Patients were encouraged, but not required, to receive post-chemoradiation chemotherapy consisting of gemcitabine at 1000 mg/m2 on days 1, 8, and 15 of each 28-day cycle for a total of 4 cycles starting 4 weeks after completing proton therapy with oral capecitabine.
Twelve patients were enrolled, but only 11 patients are reported in this analysis because 1 patient died of a gunshot wound 5 days after starting treatment. The median follow-up from the start of treatment for all patients was 14 months (range, 5 to 25 months). The median follow-up for surviving patients was 23 months (range, 8 to 25 months). The median age of enrolled patients was 68 years (range, 51 to 86 years). In total, 6 patients were female, 5 male, 2 Hispanic, and 9 white. Only 1 of the 11 patients evaluated underwent bypass surgery prior to chemoradiation. There were no interruptions in treatment for any of the patients. Treatment duration ranged from 45 to 50 calendar days. There were no interruptions in chemotherapy delivery and there were no chemotherapy-related toxicities. Seven of the 11 patients received all prescribed 66 chemotherapy doses. Two patients received 65 of the prescribed 66 doses. Two patients with insurance coverage issues received 62 of the prescribed 66 doses. Cancer antigen 19-9 levels prior to protocol treatment ranged from 13 to 3028.
Eight of the 11 patients received cytotoxic chemotherapy prior to proton therapy. Five were treated with single-agent gemcitabine and 3 were treated with the folfirinox regimen. The pretreatment characteristics of the evaluable patients, including criteria for unresectability, are shown in Table 1. The 1- and 2-year overall survival rates were 61% and 31% (median, 18.4 months; Figure 2). The 1- and 2-year progression-free survival rates were 55% and 14%. The 1- and 2-year freedom from distant metastasis rates were 68% and 27%. The 1- and 2-year freedom from local progression rates were 86% and 69% (Figure 2). No patient experienced any grade 3 or greater toxicity during treatment or during the follow-up period. Grade 2 toxicity was limited to a single patient experiencing grade 2 fatigue. Median weight loss over the course of treatment was 1.7 kg (range, loss of 5.7 to gain of 4.9 kg).
Four patients were deemed to have had an adequate radiographic response to radiation therapy to justify surgical exploration. Two of these patients underwent palliative R2 resections. Both are alive without evidence of local disease progression at 25 (Patient #5) and 22 (Patient #10) months after proton therapy, although the first developed an isolated liver metastasis and was treated with palliative stereotactic radiation therapy. The second patient developed diffuse intraperitoneal dissemination and is receiving chemotherapy. One patient (#6) underwent negative biopsies of the pancreatic head with distal pancreatectomy and was found to be without pathologically identifiable disease in any of the resected specimens. He is alive without disease progression at 24 months. One patient (#8) was found to have intra-abdominal metastases at surgery and the planned Whipple procedure was aborted.
The long-term survival expectation for a patient diagnosed with locally advanced unresectable pancreatic adenocarcinoma, even in the absence of radiographic evidence of hematogenous dissemination, remains poor. The current series is no exception with only 1 patient without evidence of disease progression at 24 months of follow-up. At the same time, the moderately high rate of local control (64% freedom from local progression at 2 years) and survival (median 18.4 months) combined with the lack of any meaningful toxicity with the regimen described herein strongly suggests the potential for treatment intensification when proton therapy is utilized in this setting.
Various studies in the contemporary radiation therapy literature discuss the utilization of stereotactic body radiation therapy in the treatment of patients with locally advanced, unresectable pancreatic cancers. While the local control rates for this modality are impressive, ranging from 84% to 94% at 12 months [12, 13], grade 3 gastrointestinal toxicity has been reported to be as high as 10% . Importantly, while the dose/fractionation regimens utilized in stereotactic body radiation therapy regimens may be effective and safe when small volumes are treated, it is unlikely that stereotactic body radiation therapy would have any relevance in the neoadjuvant setting where the value of radiation therapy would better be defined by its ability to sterilize microscopic nodal or peripancreatic disease at the periphery of the surgical field. In contrast, a prior publication from our institution  indicates that proton therapy, by virtue of the dose distributions achieved, would allow for elective nodal coverage in the neoadjuvant setting without increasing critical normal tissue exposures when compared to treatment of the gross disease alone.
With regard to normal tissue toxicity, a prior study from the M.D. Anderson Cancer Center (Houston, TX)  suggests that the duodenum is not able to tolerate doses greater than 55 Gy. This stands in contrast to the current series where, based on retrospective analysis of dose distributions, 100% of the duodenum appeared to receive the entire prescription dose of 59.40 Gy (RBE). Admittedly, 2 of the 3 patients who experienced long-term survival ultimately underwent a Whipple procedure, complicating longitudinal assessment of duodenal toxicity since their duodenums were resected. On the other hand, the absence of any suggestion of gastrointestinal toxicity in the other patients, including the 1 patient who is a long-term survivor at 2 years, raises the possibility that the duodenum may be able to tolerate higher radiation therapy doses provided other upper gastrointestinal structures (such as the jejunum and stomach) receive only minimal exposure as is the case with proton therapy.
In contrast to the current series demonstrating negligible toxicity for patients with locally advanced pancreas cancer treated with protons, Takatori et al  reported an analysis of the upper gastrointestinal complications associated with gemcitabine concurrent proton radiation therapy for patients with inoperable pancreatic cancer. The study demonstrated a 49.4% rate of gastric/duodenal ulceration in the treated patients. Terashima had previously reported favorable local control and 1-year survival rates with this regimen . Patients on this study were treated with a very aggressive radiation therapy protocol consisting of 67.5 Gy (RBE) at 2.7 Gy (RBE) per fraction with concomitant full-dose gemcitabine at 800 mg/m2 on days 1, 8, and 15. Additionally, the radiation therapy fields covered gross disease as well as regional nodal beds. While the results of this study imply that proton radiation therapy may be excessively toxic for patients with inoperable pancreatic cancer, it appears more likely that the severe toxicity demonstrated was a function of the aggressiveness of the chemoradiation therapy regimen utilized. A letter  pointing out these issues has been published.
Advanced immobilization techniques, such as the use of breath hold gating or targeting with implanted fiducial markers, were not used in this series. These technologies may be relevant for future studies with intensified treatment regimens. For the most part, motion studies were made on our 4-dimensional computed tomography scans prior to treatment delivery. The motion identified was virtually always in the range of less than 8 mm, which was well within our established planned target volume expansion beyond the IGTV. As such, it was argued that the use of breath hold or fiducial markers was unnecessary. An additional concern with shifting the patient based on fiducial marker location was the possibility that such a shift might place a critical normal structure (i.e. the spinal cord) within a high-dose volume increasing the risk of major toxicity. It will be our intention however to consider utilizing these targeting technologies in future investigations.
Proton therapy with concomitant capecitabine as delivered on the University of Florida Proton Therapy Institute PC01 protocol was well-tolerated with no grade 2 or greater gastrointestinal toxicities. Four of the 11 enrolled patients achieved a radiographic response allowing for attempted surgical resection. The median survival of 18.4 months and 2-year freedom from local progression rate of 64% compare favorably with previously reported treatment regimens. The lack of meaningful toxicity on this protocol suggests that protons may be a useful vehicle for treatment intensification in this setting. Such intensification might include the following: radiation therapy dose escalation; concomitant use of highly radiosensitizing chemotherapy regimens; hypofractionation of radiation therapy doses; or expansion of the radiation therapy treatment volume to cover likely sites of microscopic disease extensions. Finally, the favorable toxicity profile with this moderately aggressive regimen might suggest that protons would represent an optimal vehicle for radiation therapy delivery in the setting of neoadjuvant radiation therapy for patients with resectable or marginally resectable disease.
ADDITIONAL INFORMATION AND DECLARATIONS
Conflicts of Interest Disclosure: The authors have no conflicts of interest to disclose.