Consolidatory ablative stereotactic body radiation therapy after induction chemotherapy for unresectable pancreatic cancer: A single center experience

Background and purpose Consolidatory radiotherapy in form of stereotactic body radiation therapy (SBRT) with an ablative dose following induction chemotherapy is emerging as a promising treatment scheme for unresectable pancreatic cancer. Outcomes of given treatment at a single center for contiguous patients with unresectable pancreatic cancer were evaluated to build the optimal treatment strategy. Materials and methods In this retrospective study, a total of 50 patients with unresectable pancreatic cancer who underwent induction chemotherapy and ablative dose SBRT were included. SBRT dose was 40–50 Gy in five fractions. Two strategies were adopted to adhere to the organs at risk (OAR) dose constraints: simultaneous integrated protection (SIP) technique and magnetic resonance (MR)-guided adaptive technique. Overall survival (OS) and local progression-free survival (LPFS) were calculated from the start date of SBRT. Results The median follow-up period for survivors was 21.1 months (range, 6.2–61.0 months). Eleven (22.0%) patients underwent resection after SBRT, which were all R0 resection. In patients with non-metastatic disease, the median OS was 26.5 months (range, 4.1–61.0 months), and the 1- and 3-year LPFS were 90.0% (95% confidence interval [CI], 72.0–96.7%) and 57.4% (95% CI, 31.7–76.4%), respectively. Patients with oligometastatic disease had inferior survival outcomes, but there was no survival difference among responders to induction chemotherapy. In the multivariable analysis, tumor size ≤4 cm, non-metastatic status, and good response to induction chemotherapy were associated with improved LPFS. In dosimetric analysis, GTV Dmin ≥50.5 Gy was the strongest prognosticator against local progression. Grade ≥3 adverse events occurred in two (4.0%) patients with non-adaptive RT, but none in patients with MR-guided adaptive RT. Conclusion Ablative dose SBRT following induction chemotherapy is an effective strategy for selected patients with unresectable pancreatic cancer. The SIP technique and MR-guided adaptive RT were attributed to minimizing the risk of adverse events. Further studies are needed to identify the best candidates for consolidatory SBRT in unresectable pancreatic cancer.


Introduction
Pancreatic cancer is one of the leading causes of cancerrelated mortality worldwide. Most patients with pancreatic cancer present with locally advanced or metastatic disease and are not amenable to curative surgery (1). Chemotherapy is considered the standard of care for these patients, but the prognosis remains dismal, with a 5-year life expectancy of less than 10% (2). Recently, new multi-agent chemotherapy regimens have changed this paradigm (3)(4)(5). The combination regimen of 5-fluorouracil, leucovorin, irinotecan, and oxaliplatin (FOLFIRINOX) and gemcitabine with nanoparticle albuminbound-paclitaxel (Gem/nab-paclitaxel) have become the firstline treatment options for unresectable pancreatic cancer. These regimens have been demonstrated to almost double the survival of unresectable pancreatic cancer compared with the previous monochemotherapy regimens. However, the limited gain in terms of local control may possibly expand the role of RT for local control.
Stereotactic body radiation therapy (SBRT) has emerged as an attractive approach in multimodality treatment for pancreatic cancer (6)(7)(8). SBRT offers the ability to deliver a large biologically effective dose (BED) in a highly conformal manner. Furthermore, SBRT interferes less with systemic therapy as it requires only 1-2 weeks for delivery. The optimal SBRT scheme has yet to be determined, but the administration of a higher BED is essential to achieve durable tumor control and has a significant impact on survival (9). However, SBRT for pancreatic cancer is challenging due to the proximity of radiosensitive organs-at-risk (OAR), such as the duodenum, stomach, and bowel. Although prospective data regarding pancreatic SBRT is accumulating, the optimal treatment strategy remains controversial (10).
In this study, simultaneous integrated protection (SIP) technique and magnetic resonance (MR)-guided adaptive technique are implemented as components of SBRT to deliver ablative dose SBRT while minimizing the treatment-related adverse event risk. The resultant efficacy and safety of the applied treatment is evaluated. In addition, we investigated optimal SBRT strategies, including patient selection and dosimetric parameters.

Study cohort
We retrospectively identified patients with locally advanced pancreatic adenocarcinoma, including oligometastatic disease, who received induction chemotherapy followed by ablative dose SBRT between January 2017 and September 2021. Locally advanced stage was defined as a tumor with greater than 180degree involvement of the superior mesenteric artery or celiac axis or unreconstructable involvement of the superior mesenteric vein or portal vein. Oligometastatic disease was defined as less than four metastases to a single organ. Patients were excluded if they had a prior definitive treatment history, pancreatic tumor histology other than adenocarcinoma, or double-primary malignancies. A total of 50 patients with unresectable pancreatic cancer constituted the analyzed cohort ( Figure 1). This study was approved by the Institutional Review Board (IRB no. H-2105-070-1218).

Treatment
All patients received four or more cycles of induction chemotherapy before SBRT. Induction chemotherapy consisted of FOLFIRINOX (92.0%) and Gem/nab-paclitaxel (8.0%) regimens. A 3-4 week break from induction chemotherapy was required before SBRT delivery.
SBRT was performed in two ways; MR-guided adaptive SBRT was delivered using the MRIdian (ViewRay Inc., Oakwood Village, OH) and non-adaptive linear accelerator (linac)-based SBRT was delivered using the TrueBeam-STX (Varian Medical Systems, Palo Alto, CA). Patients underwent both MR and computed tomography (CT) simulations on the same day. Patients were immobilized in the supine position with arms over head. A pneumatic abdominal compressor was used to reduce breathing-induced internal tumor movement. Patients who were unable to breathe regularly, could not tolerate extended treatment times, or had any contraindications to MR (e.g., claustrophobic, metal implant) were assigned to non-adaptive SBRT. Otherwise, both adaptive and non-adaptive SBRT plans were constructed, and the best plan for an individual patient was chosen based on both OAR dose and target coverage.
The principle of tumor delineation was same for both set-up ( Figure 2). Gross tumor volume (GTV) was defined as pancreatic tumor based on 4-dimensional-CT and MR images. Clinical target volume (CTV) encompassed the GTV and vascular involvement, including the entire tumor-vessel interface. Planning target volume (PTV) included the CTV The target volumes and isodose lines for patients prescribed 50 Gy in 5 fractions with magnetic resonance-guided adaptive radiotherapy (A), and linacbased nonadaptive radiotherapy (B). GTV, gross tumor volume; PTV, planning target volume; OAR, organ at risk; PRV, planning OAR volume. Consort diagram. MR, magnetic resonance; SBRT, stereotactic body radiotherapy; OAR, organs at risk. plus a 6 mm margin. Planning OAR volume (PRV) was defined as OAR plus a 4 mm margin. If the patient had an overlapping area between PTV and PRV, PTV was divided into two; PTV sip was the overlap area between PTV and PRV, and PTV tumor was the remaining, non-overlapping PTV area with PRV. Using the SIP technique, we prescribed 33 Gy to the PTV sip , while simultaneously delivering a 50 Gy to the PTV tumor . Exceptionally, GTV was prescribed to 50 Gy even if there was an overlapping area with PRV as long as OAR constraints, which was <33 Gy to 1cc, were met. Prescribed doses were lowered down to 40 Gy depending on the situation, including target/ organ movement and target to organ distance of the individual patient. SBRT was delivered in five fractions with the goal of 95% PTV coverage with 100% of the prescription dose, prioritizing hard OAR constraints. OAR dose constraints were as follows: for the duodenum, bowel, and stomach, Dmax <35 Gy and D1cc <33 Gy; for the spinal cord, Dmax <22 Gy; for kidneys, Dmean <10 Gy; and for the liver, 700 cm3 <21 Gy. If the OARto-target distance was sufficient and the PTV coverage and OAR constraints were simultaneously met, the patient was often assigned to non-adaptive linac-based SBRT, due to the convenience of a shorter treatment delivery time. In contrast, if PTV coverage could not be met due to a violation of OAR constraints, the patient was usually assigned to MR-guided adaptive SBRT.
For adaptive SBRT delivery, MR imaging was performed before each fraction, and new OARs were re-contoured, reflecting inter-fractional changes. Then, the new re-optimized plan was generated. If the new plan violated OAR constraints, plan normalization was altered till OAR constraints were fulfilled. After the daily adaptive plan was determined, each fraction was delivered under MR-guided real-time gating. In non-adaptive SBRT, image-guided RT using cone beam CT was performed for all fractions. To reduce inter-and intra-fractional variability, a 6-hour fasting period before simulation and each fraction was mandated.

Follow-up
After completing SBRT, physical examination, laboratory tests, and imaging were performed every 2-6 months or when clinically indicated. Most patients continued to receive additional systemic therapy. The multidisciplinary tumor board determined resectability after reviewing the imaging and performance status. Adverse events were graded using the Common Terminology Criteria for Adverse Events (version 5.0). Acute adverse events were defined as those occurring within three months after SBRT, whereas late adverse events were defined as those occurring beyond three months after SBRT.

Statistical analysis
Response evaluation was performed using response evaluation criteria in solid tumors (RECIST). The response to induction chemotherapy was evaluated in two ways. The overall response was measured by comparing the disease status between the time of initial diagnosis and the time of SBRT. Whereas, the response at SBRT referral compared disease status between three months before SBRT and at the time of SBRT. For dosimetric analysis, all parameters were depicted from the cumulative daily delivered re-optimized plans for the adaptive treatments, whereas the original treatment plan was used for the non-adaptive treatments. GTV Dmean was defined as the mean dose absorbed by the GTV. GTV Dmin and GTV Dmax were defined as the minimum and maximum dose absorbed by 1 cc of GTV, respectively.
Local progression was defined as a 20% or more increase in tumor size on the CT scan compared to previous imaging following the RECIST criteria. Local progression-free survival (LPFS) was calculated from the start date of SBRT to the date of local progression or death. Overall survival (OS) was calculated from the start date of SBRT to the date of death. Surviving patients were censored at the date of the last follow-up. Survival outcomes were estimated using the Kaplan-Meier method. Cox proportional hazards modeling assessed whether survival outcomes varied according to risk factors. Continuous variables, including dose parameters, were divided into two subgroups at cutoff values identified by receiver operating characteristic (ROC) curves and then analyzed. All statistical tests were performed using STATA (version 15.1; StataCorp LP, College Station, TX, USA).
A dramatic decline in carbohydrate antigen (CA) 19-9 was observed after induction chemotherapy and SBRT. The median % change of CA 19-9 was -82.7% from baseline to postinduction chemotherapy/pre-SBRT and -38.8% from postinduction chemotherapy/pre-SBRT to post-SBRT. GTV Dmin, the minimum dose absorbed by 1cc of the GTV. 2 GTV Dmax, the maximum dose absorbed by 1cc of the GTV. 3 GTV Dmean, mean dose absorbed by the GTV.

Prognostic factors
The univariate and multivariate Cox proportional hazards models for OS and LPFS are presented in Table 2. In the multivariate analysis, M-stage was a strong prognostic factor for both OS and LPFS (Hazard ratio [HR] 2.74, p=0.018 in OS; HR 5.32, p=0.003 in LPFS). A large initial tumor size (≥4 cm) was associated with inferior LPFS (HR 4.26, p=0.003). Response to the induction chemotherapy in overall (HR 3.79, p=0.044) and at SBRT referral (HR 4.27, p=0.001) were both significantly related to OS, but only response at SBRT referral was a significant factor for LPFS (HR 3.72, p=0.022). Among dose parameters, higher GTV Dmin (≥50.5 Gy) was significantly associated with improved LPFS (HR 3.06, p=0.045) and OS (HR 2.58, p=0.031). However, the prescribed dose (≥50 Gy), GTV Dmax (≥52 Gy), and GTV Dmean (≥51 Gy) failed to demonstrate statistical significance. Greater CA 19-9 decline after SBRT (≥50%) had a strong relationship with improved LPFS (HR 3.85, p=0.015). On the other hand, the volume of PTV and PTV sip did not affect survival outcomes. SBRT technique (adaptive vs. non-adaptive) also had no significant association with both OS and LPFS.

FIGURE 3
Kaplan-Meier curves of (A) overall survival, and (B) local progression-free survival according to metastatic status. SBRT, stereotactic body radiotherapy.

Dosimetric analysis
The values of dosimetric parameters in patients with or without local progression within three years following SBRT are shown in Table 3. The mean values were all higher in patients without local progression compared to those with local progression. Among dosimetric parameters, the difference between the two groups was significant for the prescribed dose  (Figure 4).

Adverse events
The acute and late adverse events are presented in Table 4. Acute grade 2 adverse events occurred in 23 (46.0%) patients. There were no grade 3 or higher acute adverse events. The most common acute adverse events were abdominal pain (32.0%), nausea/vomiting (28.0%), and poor oral intake (16.0%). Regarding the late adverse events, grade 2 or higher events occurred in 15 (30.0%) patients. A total of two (6.0%) patients experienced grade 3 gastro-intestinal bleeding, both requiring endoscopic intervention, but were successfully managed. There were no grade 4 or higher late adverse events. Patients treated with MR-guided adaptive SBRT showed a trend toward lower rates of late adverse events compared with those treated with non-adaptive SBRT (grade 2: 19.2% vs. 33.3%; grade 3: 0% vs. 8.3%; p=0.084).

Discussion
Pancreatic cancer is perceived as a systemic disease with the eventual emergence of widespread metastases. Despite this, up to 30% of patients with pancreatic cancer are reported to be without metastatic disease at the time of death (11). This percentage is likely to further increase as systemic therapy is continuously advancing in both multiagent cytotoxic approaches and combined precision medicine-based strategies (4,12). In an era of more effective systemic therapy, maximizing local treatment has become more important, which may eventually lead to improved treatment outcomes. However, the use of conventionally fractionated RT failed to convey survival benefits in addition to the standard of care chemotherapy in a prospective trial, despite improvement in local control (13). SBRT, stereotactic body radiotherapy; GTV, gross tumor volume; AUC, area under curve. GTV Dmin, the minimum dose absorbed by 1cc of the GTV; GTV Dmax, the maximum dose absorbed by 1cc of the GTV; GTV Dmean, mean dose absorbed by the GTV. *Data are expressed as mean ± standard deviation.
In this context, SBRT has emerged as an attractive alternative showing the potential to improve local control, with an ability to deliver a higher dose in a conformal manner and requiring shorter overall treatment time, simultaneously. A recent meta-analysis reported that SBRT significantly improved 2-year OS compared to conventionally fractionated RT (26.9% vs. 13.7%, p = 0.004) (6). Petrelli et al. performed a systemic review of 19 trials of SBRT and reported the pooled 1-year OS rate of 51.6% (95% CI 41.4-61.7%) and 1-year local control rate of 72.3% (95% CI 58.5-79%) (14). Many of these data suggest that a higher dose is needed to achieve adequate tumor control (15,16). Toesca et al. reviewed the treatment outcomes of 149 patients who received multi-fraction SBRT for unresectable pancreatic cancer (15). They reported that patients treated with SBRT dose ≥40 Gy had superior OS and PFS compared to those who received SBRT dose <40 Gy (median OS: 23 vs. 14 months, p=0.0007; median PFS: 13 vs. 10 months, p = 0.007). In this study, we reported the 1-year OS and LPFS rates of 87.3% and 90.0%, respectively, which is in line with SBRT series with an ablative-dose (BED 10 ≥100 Gy). However, SBRT delivering an ablative dose is still limited due to the proximity of critical neighboring radiosensitive OARs. Furthermore, the tumor and surrounding structures are highly mobile and sometimes difficult to identify using cone beam CT-based imaging. Thus, in this study, we adopted two approaches to safely deliver ablative doses.
The SIP technique allows the simultaneous delivery of ablative doses to the tumor volume, whereas the overlapping volume with critical OARs is covered by a lowered, safer dose (17). Several studies have reported the results of SIP protocol in different RT schemes (18-20). Simoni et al. performed SBRT using the SIP technique by administering 50 Gy to the tumorvessel interface (TVI), 30 Gy to the pancreatic tumor, and 25 Gy to the SIP volume (18). They found no acute or late grade ≥3 adverse events, but a predominant incidence of in-field failures occurred. After that, the authors investigated dose escalation protocols up to 60, 40, and 33 Gy to TVI, pancreatic tumor, and SIP volume, respectively, and demonstrated the feasibility with adequate PTV coverage and acceptable OAR exposure (21). In this cohort, we prescribed 50 Gy to the TVI and tumor and 33 Gy to the SIP volume. We found favorable toxicity profiles without compromising survival outcomes. The volume of PTV and PTV sip did not affect survival outcomes in multivariable analysis, although the small cohort size may have obscured the impact.
On the other hand, MR-guided adaptive RT is another powerful tool ensuring accurate treatment delivery with several advantages (22). First, MR guidance offers improved soft tissue contrast, resulting in the ability to distinguish the boundaries of pancreatic tumors. Second, a new re-optimized plan can be generated and optimized per fraction, reflecting daily anatomical changes. Hassanzadeh et al. found that duodenal dose constraints would have been violated in 67.7% of fractions for pancreatic cancer patients without per fraction optimization (23). Third, a real-time gating system enables intra-treatment monitoring of OARs. Several retrospective studies of MR-guided adaptive RT have reported promising outcomes while minimizing toxicities (24,25). In the current study, grade ≥ 3 adverse events occurred in two (4.0%) patients with nonadaptive RT, and none in patients with MR-guided adaptive RT. It is notable that patients with close OAR-to-target distance were mostly assigned to MR-guided adaptive SBRT (Figure 1). Therefore, MR-guided adaptive RT may be more suitable for patients with a high risk of adverse events. Given the low conversion rate to surgery, safety concerns should be a top priority for SBRT treatment. The adoption of SIP technique and MR-guided adaptive RT can contribute to minimizing the risk of adverse events while delivering ablative doses to the tumor. Currently, the optimal SBRT strategies have yet to be determined for pancreatic cancer. Published studies have used various combinations of dose and fraction schemes. Nonetheless, in order to perform SBRT effectively, particularly in daily adaptive SBRT, it is essential to identify significant dose parameters. Current dosimetric analysis nominated GTV Dmin as the most relevant parameter against local progression, with the cutoff value of 50.5 Gy, as determined by ROC analysis. Thus, patients with smaller PTV and PTV sip volumes may be better candidates for SBRT that meet the dosimetric requirements and minimize adverse events. In addition, our results found that patients who had a tumor size ≤4 cm, non-metastatic status, and good response to induction chemotherapy showed a better prognosis after SBRT. Response to induction chemotherapy at three months prior to SBRT was more predictive of treatment outcomes compared to at the time of diagnosis. This may reflect that prompt delivery of consolidatory treatment would be a more effective strategy for patients with an initial response to induction chemotherapy but a stationary response afterward, compared with continued chemotherapy till progression.
Furthermore, it is notable that patients who had oligometastatic disease but achieved response to induction chemotherapy showed similar survival rates compared to nonmetastatic patients in this study. Only few studies have examined the role of SBRT in metastatic pancreatic cancer. Lischalk et al. evaluated 20 patients with metastatic pancreatic cancer who received chemotherapy and SBRT (26). They reported 1-year OS and local control rates of 43% and 53%, respectively, without grade ≥3 toxicities. They also found that smaller PTV was associated with improved OS (p=0.001) and local control rates (p=0.02). Rosati et al. recommended a minimum of 6 months of chemotherapy and an observation period (4-8 weeks) before SBRT to better understand the natural history of disease (27). In this study, we found that a longer duration of induction chemotherapy (≥3 months) and good response to chemotherapy were associated with improved survival outcomes. Although based on the observation from very selected small population, SBRT to the primary site may have a role for oligometastatic patients who underwent chemotherapy over a period of time and achieve a durable response to induction chemotherapy. On the contrary, we do not advocate SBRT for patients with oligometastatic disease who did not respond to induction chemotherapy, as patients with these unfavorable factors eventually experienced disease progression within six months from SBRT. This principle is in line with selectively offering ablative local treatment in oligopersistent disease.
There are several limitations to this study. First, the number of patients is far insufficient to draw concrete conclusion, especially in the oligometastatic disease subgroup. Additionally, as this was a retrospective study, the events could have been underestimated due to incomplete medical records. Second, the dosimetric analysis might be biased because lower doses were prescribed in patients with poor performance status. Finally, MR-guided adaptive SBRT was performed using the Cobalt-60 system in this study. Better OAR sparing may be achieved with MR-linac system with a steeper dose gradient. Nonetheless, the entire cohort was treated with the uniform treatment protocol, including dose prescription, target delineation, and decision policy for radiotherapy technique. Furthermore, they all received FOLFIRINOX or Gem/nab-paclitaxel, which are currently considered the standard of care as the first-line treatments. Thus, our findings may have better applicability and generalizability to current clinical practice.
In conclusion, consolidatory ablative dose SBRT following induction chemotherapy could be a viable treatment option for selected patients with unresectable pancreatic cancer. Patients with a tumor size ≤4 cm and achieved a durable response to induction chemotherapy may be good candidates for SBRT treatment as consolidatory measure. Prompt ablative consolidatory treatment delivery may be more appropriate approach compared to sustained chemotherapy beyond initial response. For SBRT planning, GTV Dmin (≥50.5 Gy) was identified as the most relevant parameter against local progression. The SIP technique and MR-guided adaptive RT strategies enabled the delivery of an ablative dose to the tumor while minimizing toxicity of surrounding OARs. Further studies with larger cohort sizes, better yet prospective design would help to further validate the role of the optimal SBRT strategies in pancreatic cancer.

Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement
The studies involving human participants were reviewed and approved by The Institutional Review Board of Seoul National University Hospital (IRB no. H-2105-070-1218). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

Author contributions
EC contributed conception and design of the study. HL contributed to data collection, statistical analysis and wrote the first draft of the manuscript. EC and H-CK revised the manuscript. All authors contributed to the article and approved the submitted version.