An integrated boost with conventional fractionation to the tumor-arterial interface is unlikely to lead to downstaging in borderline resectable and locally advanced pancreatic cancer.
About the lead author:
Martin C. Tom, MD
Houston, TX 77030
Eugene J. Koay, MD, PhD
Department of Radiation Oncology
The University of Texas MD Anderson Cancer Center
Sarah Hoffe, MD
lack Department of Radiation Oncology
Moffitt Cancer Center
Why is this article contemporary?
Although homogeneous dose prescriptions have been the standard in radiation therapy delivery, advanced technology integration now creates the potential for dose painting (DP). With the article by Crane et al in this issue, differential dose deposition is explored in those pancreatic cancer patients with extensive vascular involvement on the basis that higher biologically effective doses may be needed to facilitate conversion to negative margin resection. The incorporation of intensity-modulated radiation therapy and motion management allow the highly precise delivery of 63 Gy to that portion of the tumor at higher risk for residual disease, should resection become feasible.
Contextually, this data invites consideration of novel dose painting strategies, suggesting that the future may invite innovative heterogeneous dose escalation. Recent data from radiomic studies have reported that intra-tumoral heterogeneity is present with marked molecular differences between cells in the core of the tumor vs the tumor edge.1 Analysis of metastatic liver tumor deposits from colorectal cancer has shown that those tumors with higher intra-individual genetic heterogeneity have worse 3-year survival.2 With the improved imaging techniques now available, the near future may herald the capacity for the prospective imaging delineation of clinically relevant microhabitats to guide molecularly targeted radiation therapy.
This article therefore sets the stage for new dose exploration in the setting of patients with unresectable pancreatic cancer. The authors report their experience with this technique, raising the question of how radiation oncologists can best maximize a DP approach to improve outcomes. Improved correlation of imaging data with intra-tumoral molecular phenotypes has the future potential to significantly accelerate a new dosimetric paradigm.
1. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data. Radiology. 2016;278(2):563-577. doi: 10.1148/radiol.2015151169.
2. Sveen A, Loes IM, Alagaratnam S, et al. Intra-individual genetic heterogeneity among liver metastases in metastatic colorectal cancer. ASCO Meeting Abstracts. 2016;34(4_suppl):555.
Houston, TX 77030
There is substantial interest in the use of escalated doses of radiation to improve tumor resectability in pancreatic cancer, but true radiographic downstaging does not frequently occur. We reviewed our experience integrating a radiation boost to the tumor-arterial interface in borderline resectable and locally advanced pancreatic cancer with the objective of downstaging.Twenty-three patients with borderline resectable (n = 14) or locally advanced disease (n = 9) received chemoradiation with an integrated radiation boost to the tumor-arterial interface (median dose: 63 Gy in 28 fractions). We recorded patient outcomes and compared CT scans before and after radiotherapy to evaluate response.There was no downstaging in 21 of 23 patients. The 2 patients who achieved downstaging of their primary tumor had early distant metastasis (DM) that precluded surgery. Three patients with borderline resectable disease underwent margin-negative resection. Many patients developed DM soon after chemoradiation as median distant metastasis-free survival (DMFS) was 8 months. The median local progression-free survival (LPFS) was 23 months and median overall survival (OS) was 16 months.An integrated boost with conventional fractionation to the tumor-arterial interface is unlikely to lead to downstaging in borderline resectable and locally advanced pancreatic cancer.
Surgical resection remains the only potentially curative treatment for pancreatic cancer. However, fewer than 20% of patients present with potentially resectable disease.1 Consequently, outcomes remain poor with an estimated 5-year relative survival rate of 6%.2 For those patients who proceed to surgical resection, negative margins (R0) have been established as crucial in attaining long-term survival.3 Pancreatic cancer has a tendency to spread along the vasculature of the retroperitoneal border, and the most common site of microscopically positive resection is the superior mesenteric artery (SMA) margin.4,5 Thus, in the absence of distant metastasis (DM), the primary determinant of resectability in pancreatic cancer is the extent of vascular involvement.
Because the majority of patients present with locally advanced disease, significant attention has been focused on neoadjuvant (preoperative) strategies to decrease the size and extent of the primary tumor, thus increasing the probability of R0 margins in patients with borderline resectable and, to a lesser extent, locally advanced tumors. The definition of borderline resectable includes a very heterogeneous group ranging from minimal venous involvement (T3) to extensive arterial involvement with or without venous involvement (T4).6-8 Rigidly defined, downstaging in pancreatic cancer is characterized as changing the radiographic stage from T4 to T3 with the use of neoadjuvant therapy. The term “downstaging” can also be used to describe a reduction of the amount of arterial involvement (T4) to less than 180 degrees that would allow surgical resection. However, it is often incorrectly associated with T3 tumors with venous involvement that eventually undergo successful resection after neoadjuvant therapy.
Methods and MaterialsPatients
Downstaging following conventional doses of neoadjuvant chemoradiation is a rare occurrence in patients with borderline resectable disease.9 However, recent studies have suggested administering higher doses of neoadjuvant radiation can improve downstaging of pancreatic cancer.10,11 Still, data are inconsistent with regard to this phenomenon and are further confounded by varying definitions and subjective criteria of resectability among various trials and institutions.6,7 Here, we reviewed our institutional experience of patients with borderline resectable and locally advanced pancreatic cancer with extensive venous or arterial involvement who received neoadjuvant chemoradiation with a strategy of dose escalation to the tumor-arterial interface with the objective of downstaging.The University of Texas MD Anderson Cancer Center (MDACC) Institutional Review Board approved this study. We retrospectively reviewed all patients who were treated with neoadjuvant chemoradiation with dose escalation to the tumor-arterial interface at MDACC between 2005 and 2010. We identified 23 patients and reviewed their characteristics, treatment regimens, and outcomes.
Patient and Tumor Characteristics
Based on vessel involvement, patients were classified as potentially resectable, borderline resectable, or locally advanced by MD Anderson Criteria and the definition proposed by the American Hepatopancreatobiliary Association (AHPBA), Society of Surgical Oncology (SSO), and Society for Surgery of the Alimentary Tract (SSAT) and endorsed by the National Comprehensive Cancer Network (NCCN).8,12,13 The primary tumor stage (T stage) was determined by American Joint Committee on Cancer (AJCC) 7th edition guidelines.Characteristics of the 23 patients are shown in Table 1.Twenty-two patients received gemcitabine-based induction chemotherapy for a median duration of 3 months (range, 1-9 months) and underwent subsequent intensity modulated radiotherapy (IMRT) with concurrent chemotherapy.
Radiologic Review and Classification of Resectability
Image guidance was achieved using daily orthogonal film alignment. Primary tumor and nodal regions were treated to 50.4 Gy in 28 fractions. The median escalated dose to the tumor-arterial interface was 63 Gy (range, 57.5-64.4 Gy) in 28 fractions (range, 25-35 fractions). Assuming an alpha/beta ratio of 10, median BED10 was 77 Gy. An example of the integrated boost technique is shown in Figure 1. Eight patients proceeded to laparoscopy or laparotomy after neoadjuvant therapy to evaluate for resection. Four patients underwent resection.Bhosale reviewed the pre-therapy and immediate post-chemoradiation staging CT scans for all patients. We recorded tumor size, regional node status by RECIST criteria (>15mm), and tumor involvement of the SMA, celiac axis (CA), common hepatic artery (CHA), superior mesenteric vein (SMV), portal vein (PV), and superior mesenteric vein-portal vein confluence (SMV-PV). Tumor-vessel involvement was documented as no involvement, abutment (<180 degrees of vessel circumference), or encasement (>180 degrees of vessel circumference). Occlusion of the SMV-PV was also recorded. To characterize the changes in tumor size and disease stage, Bhosale compared the greatest dimension of the primary tumor and each tumor-vessel interface depicted in pre-treatment and post-chemoradiation images at first restaging. Changes were described using definitions in Supplemental Table 1 and by modified RECIST criteria (version 1.1).14 To grade response of the primary tumor alone, we also report RECIST classifications with omission of DM as criteria for progressive disease (PD).
Downstaging was defined as radiographic conversion of a T4 to a T3 stage tumor. Additionally, our definition included conversion of the tumor stage from locally advanced to borderline resectable disease, or borderline resectable to potentially resectable disease by MD Anderson and/or AHPBA/ SSO/SSAT criteria. Tumor-vessel improvement was defined as conversion from encasement to abutment, encasement to no involvement, or abutment to no involvement. Tumor- vessel worsening was defined as conversion from no involvement to abutment, no involvement to encasement, or abutment to encasement. Additionally, for patients who progressed locally, we compared the CT scan demonstrating local progression with treatment planning images to determine the local pattern of failure by isodose line.Duodenal toxicity was recorded and scored according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0. All of the recoded duodenal toxicity events were confirmed by endoscopic evaluation. The time to toxicity was defined in months after the completion of radiation therapy to endoscopic confirmation.We recorded failure patterns at any time after treatment and did not censor patients after first evidence of progression. Local progression free survival (LPFS) was defined as the time from beginning of induction chemotherapy to primary tumor PD on imaging or surgical exploration; we censored patients who died without evidence of primary tumor PD. Distant metastasis free survival (DMFS) was defined as the time interval between beginning of induction chemotherapy and development of DM found by either imaging or surgical exploration; we censored patients who died without evidence of DM. Overall survival (OS) was defined as the time from initiation of induction chemotherapy to death.Inability to Resect Due to Progressive Disease Of the 14 patients who were initially borderline resectable, 8 patients showed radiographic progression which precluded surgery (1 progressed locally and 7 developed DM).
The other 6 patients with borderline resectable disease underwent surgical exploration. Three of these patients were found to have unresectable disease based on DM (n=2) or local tumor anatomy rendering the patient unresectable (n=1). The other 3 patients who underwent surgical exploration had R0 resections; none of these 3 patients had any major unexpected acute or long-term complications after escalated radiation and surgery.
Lack of Radiographic Downstaging After Dose Escalation
Of the 9 patients with initially locally advanced disease, 7 patients remained unresectable. One patient was found to have DM at laparoscopy. Another patient was deemed unresectable, but did not follow up with surgeons at MDACC and underwent resection at an outside institution.All 23 patients were restaged after chemoradiation (median duration of 1.3 months, range of 0.6-1.5 months). Three of the 23 patients met radiographic criteria to change the stage of their primary tumor, 2 of which were downstaged while the third was upstaged. Both patients whose primary tumors were downstaged developed DM, which precluded surgical resection.
Local Pattern of Failure
By RECIST criteria, 10 patients (43%) had stable disease (SD), 5 patients (22%) had a partial response (PR), and 8 patients (35%) had PD because of an increase in the greatest dimension of the primary tumor (n = 1) or the development of DM (n = 7). When analyzing only the primary tumor and omitting DM as criteria for PD, 15 patients (65%) had SD, 7 patients (30%) had a PR, and 1 patient (4%) had PD (Figure 2). No patients had a complete response. Of 72 tumor-involved vessels, following treatment, 9 (13%) demonstrated radiographic improvement and 5 (7%) worsened. Detailed response of the tumor- vessel interface is reported in Supplemental Table 2.Of the 8 patients who progressed locally, 5 patients progressed within the 50.4 Gy isodose line, and 3 patients progressed within the 54 Gy isodose line. Eighteen of the 23 patients in this study were included in a recent analysis of duodenal toxicity following chemoradiation at our institution.15 In the current study, 5 of the 23 patients (22%) were noted to have duodenal toxicity occurring at a median of 3.4 months (range, 3-11) after treatment. One patient had grade 1 duodenitis, which was found incidentally. One patient had a grade 2 bleeding duodenal ulcer. Two patients had grade 3 bleeding duodenal ulcers requiring blood transfusion. One patient developed grade 5 toxicity. This patient was found to have a large bleeding duodenal ulcer at 11 months after radiation that was not controlled by argon plasma coagulation. Given the patient’s extensive disease progression, the family opted for comfort care and the patient died. No patients experienced duodenal perforation. Median follow-up was at 17 months. All 23 patients died, and median OS was 16 months. Eight patients had local progression. The median LPFS was 23 months. Twenty patients developed DM. The median DMFS was 8 months. Kaplan Meier estimates of OS, LPFS, and DMFS are shown in Figure 3.In this study we provide a detailed analysis of patients with borderline resectable and locally advanced pancreatic cancer who underwent neoadjuvant chemoradiation with an integrated radiation boost to the tumor-arterial interface with the objective of downstaging. Following dose escalation to a median of 63 Gy in 28 fractions (BED10=77 Gy) to this region, we did not observe clinically significant downstaging. Eight patients underwent laparoscopy or laparotomy to evaluate for resection after neoadjuvant therapy, but only 4 patients underwent surgical resection (one at another institution). Notably, 8 patients progressed locally within the 54 Gy isodose line or below. Furthermore, potential for duodenal toxicity is a substantial concern with dose escalation efforts, and incidence of toxicity demonstrated a dose-volume relationship. Collectively, our data caution against aggressive efforts to downstage locally advanced or borderline resectable tumors by escalating radiation dose solely to the tumor-arterial interface.
Our data add to a growing body of literature exploring this treatment strategy, and the results are mixed. Some recent studies have reported dose escalation strategies resulting in significant downstaging of borderline resectable and locally advanced pancreatic cancer. One retrospective study evaluated outcomes of patients treated with gemcitabine followed by stereotactic body radiotherapy (SBRT) to 30 Gy in 5 fractions, with dose escalation to 35 Gy in the region of vessel abutment/encasement.
Of patients with borderline resectable disease, 77% underwent surgery, with 56% completing resection (97% R0).11 The authors noted a significant radiographic response in most patients, but details of vessel involvement after treatment were not reported. Notably, a large proportion of patients in this study reported initial borderline resectable stage T3 tumors (77%), which by definition did not involve the SMA or CA. Since the analysis did not include changes in radiographic staging, it is unclear if SBRT contributed to downstaging in this study. By comparison, our cohort included 70% of patients with T4 tumors with extensive arterial involvement, who are much less likely to undergo resection.
A recent dose escalation study using gemcitabine with concurrent IMRT for locally advanced pancreatic cancer resulted in a recommended dose of 55 Gy in 25 fractions. Although all patients were initially locally advanced, 12 of the 50 patients (24%) went on to surgical resection (83% R0).10 The authors reported significant radiographic improvement after therapy; however, details of tumor-vessel response were not provided. In our subgroup of patients with locally advanced disease, only 1 of 9 (11%) had radiographic evidence to warrant downstaging (and was later found to have DM at laparoscopy). Similarly, in the aforementioned SBRT study11, only 2 patients (12.5%) with locally advanced disease proceeded to surgery, both of whom were eventually deemed unresectable.
Dose escalation strategies have also demonstrated improved local control rates for pancreatic cancer, but have been limited by higher rates of gastrointestinal toxicity. 16-20 In the dose escalation study by Ben-Josef and associates10, 8 patients (17%) progressed locally with a 2-year freedom from local progression (FFLP) of 59%. Of the 20 patients treated to 55 Gy, 6 grade > 3 events occurred, leading to an estimated 24% probability of dose-limiting toxicity. With SBRT11, local failure only occurred in patients who did not undergo resection. Of those patients, local control was 60% at 18 months. Furthermore, only 5.3% experienced late grade 3 toxicity with no grade 4 or 5 events. In our study, 6 patients progressed locally with an estimated 2-year FFLP of 33%. We encountered an overall gross rate of 13% grade > 3 late duodenal toxicity. Interestingly, our pattern of local failure analysis revealed that no patients failed above the 54 Gy isodose line, which may inform future dose escalation strategies.
Our results indicate that the goal of escalated radiation therapy should be shifted away from improving resectability to improving local control in borderline resectable and locally advanced pancreatic cancer. More standardized resection and response criteria are necessary to better compare results within the literature. Similarly, detailed analyses of vessel involvement and local patterns of failure are helpful to further aid the choice of optimal dose, as are toxicity rates to determine the therapeutic ratio. Importantly, DM continues to play a large role in preventing surgical resection and causing death; therefore, escalated radiation doses should be combined with either improved systemic therapies or biomarkers that help predict patterns of failure.
While this study provides some guidance for future efforts of radiation therapy in pancreatic cancer, we acknowledge its limitations. Primarily, this study was retrospective and had a small sample size. Additionally, although radiand treatment courses were not uniform. Finally, there was inherent bias in the selection of the patients for dose escalation. However, it must be noted that the selected patients exhibited favorable tumor characteristics—namely, lack of progression prior to starting chemoradiation—and yet all had poor outcomes regardless.
In summary, dose escalation to 63 Gy in 28 fractions to the tumor-arterial interface resulted in low rates of tumor resection. In patients with borderline resectable disease, development of DM soon after chemoradiation was the main factor that precluded surgery. In patients with locally advanced disease, lack of tumor response was a major reason patients were not resected. For the 3 patients who underwent surgery for borderline resectable disease, margins were negative, and none experienced a major unexpected complication from the escalated dose of radiation.
Given these results, we have adopted a strategy using ablative doses to the entire tumor in select cases. To avoid the duodenal toxicity we encountered using only orthogonal kV image guidance, we now use strict mucosal dose constraints and daily CT-guided 3D-3D matching for image- guided radiation therapy.
ABOUT THE AUTHORS
Houston Methodist (MT), Houston, TX. Department of Radiation Oncology (EK, PD, CC), Department of Diagnostic Radiology (PB), Department of Surgical Oncology (MK, JF, JL), Department of Gastrointestinal Medical Oncology (GV, RW), The University of Texas MD Anderson Cancer Center, Houston, TX.
Address correspondence to: Christopher H. Crane, MD, MD Anderson Cancer Center, Department of Radiation Oncology, 1515 Holcombe Blvd. MS-97, Houston, TX 77030, Phone: 713-563-2341. Fax: 713-563-2331. E-mail: firstname.lastname@example.org
Disclosures: Dr Varadhachary participates in consulting for Celgene. Conflicts of interest: None.
Sources of funding: The authors acknowledge support from the Center for Radiation Oncology Research and the Sheikh Ahmed Center for Pancreatic Cancer Research at The University of Texas MD Anderson Cancer Center. Dr Koay was also supported by NIH’s Physical Sciences Oncology Centers (PS-OCs) U54CA143837 (Center for Transport Oncophysics), the Pancreatic Cancer Action Network (14-20-25-KOAY) and the Radiological Society of North America (RSD1429).