Pancreatic cancer remains one of the diseases with the worst prognosis in oncology . According to stage at presentation, patients are usually divided into three groups: resectable disease (~20%), locally advanced disease (~40%) and metastatic disease (~40%) . Even in the most favourable group of patients with resectable cancer, the 5-year overall survival is limited to about 20% , and the vast majority of these patients still develops an incurable treatment failure, mainly distant metastasis. However, in 10-35% of the patients an isolated local recurrence without evidence for distant spread occurs [4, 5]. Although an isolated local recurrence represents a potentially curable situation, little interest has been paid to specifically address this situation, no generally accepted standard treatment exists, and a special follow-up for patients after resection of pancreatic cancer is not recommended by international guidelines. Consequently these patients are treated within a variety of concepts initially developed for different stages of primary pancreatic cancer, ranging from aggressive surgical approaches as in primary resectable disease stages or chemoradiation as in locally advanced pancreatic cancer to palliative chemotherapy as in metastatic situations. However, there are some arguments for addressing this situation as a specific stage of disease. In our cohort, the median time to local recurrence from primary treatment was 20 months. In contrast to primary resectable pancreatic cancer, these patients did obviously not develop early distant metastasis, which is a common feature after curative intended surgery. For example, in the Conco-001 trial , which addressed the issue of adjuvant chemotherapy in resectable primary pancreatic cancer, the median disease-free survival in the experimental arm was only 13 months, mainly due to early distant failure, while local recurrence was observed only in about 20% of the patients. In contrast to metastatic disease, these patients could rationally benefit from a local treatment component as they suffer from a locoregional confined situation. Although they share some features with LAPC patients like aggressive local invasion into adjacent structures, they differ to some extent from this patient group as they have already been exposed to surgery and frequently chemotherapy. Therefore, it seems reasonable to assume (at least to some extend) a different disease biology in these patients, and to specifically address this situation as a unique stage of disease.
To our knowledge, our cohort represents one of the largest series published in peer reviewed literature that specifically addresses the treatment and outcome for isolated local recurrences of pancreatic cancer. With our treatment approach which includes radical surgery, IORT and chemoradiation, we observed encouraging rates of local control, disease-free survival and overall survival with acceptable toxicity. Notably, 6 of our 36 patients lived for more than 3 years and the estimated median survival of 19 months after the treatment of local recurrence found in our analysis compares favourable with the results published by other groups addressing LAPC (which seems to be the best available benchmark in the absence of randomized trials or larger cohort series specifically addressing isolated local recurrences). Using chemotherapy alone, modern phase III trials consistently reported median overall survival rates of only 9 to 10 months [8–10] for LAPC and about 6 to 8 months for metastatic disease [10–12]. The evaluation of chemoradiation approaches within modern phase II and III trials resulted in improved locoregional control rates but only in slightly improved median overall survival rates of 11 to 12 months [8, 13–15], although some single center studies reported median overall survival rates of 17-19 months due to intensification of treatment [16–18].
Similar to primary pancreatic cancer, long term survival is hardly imaginable in patients with isolated local recurrences without the achievement of locoregional control. But even in resectable primary pancreatic cancer, local recurrences occur in up to 75% of the patients based on autopsy findings due to the always narrow and often positive margins . Adjuvant chemoradiation using fractionated external beam radiation therapy (EBRT) has proven to reduce local recurrence rates compared to chemotherapy alone regardless of its conflictingly reported influence on overall survival . However, the dose that can be safely applied with external beam radiotherapy is limited due to the tolerance of surrounding structures at risk, mainly small bowel. Even with the use of intensity-modulated therapy (IMRT), doses beyond 60 Gy, which are potentially needed to sterilize at least microscopic residual disease , can hardly be achieved without unacceptable toxicity. Especially surrounding bowel structures cannot be adaequately spared due to the narrow anatomical relations and the intra- and interfractional movements. Therefore, other radiation techniques, namely stereotactic radiosurgery (RS) and intraoperative radiation therapy (IORT) have been investigated as sole treatment or as boost techniques in combination with EBRT to achieve further dose escalation in LAPC. Notably, most of these trials also included small numbers of isolated local recurrences, but subgroup analyses were not available [21–23]. In the initial experience with RS published by the Stanford group, RS resulted in excellent short term local control rates (80-100%), but did not improve median survival (median OS 8-11 months) compared to fractionated chemoradiation and raised some concerns about late toxicity of high single doses in the pancreatic region [21, 22]. However, Mahadevan et al.  recently reported a series of 36 patients with unresectable pancreatic cancer treated with 3 fractions of stereotactic radiotherapy up to total doses of 24-36 Gy followed by gemcitabine using an adaptive tolerance-based dose prescription approach dependent on the distance of the target to duodenal structures. After a median follow-up of 24 months, a local control rate of 78% with a median disease-free survival of 9 and a median overall survival of 14 months was found .
In our study, we observed comparable results (local control 83%) using intraoperative radiation therapy (IORT) as a boosting strategy in combination with maximal surgery and moderate doses of external beam radiotherapy to overcome the dose limitations mentioned above. In contrast to all external radiation therapy techniques, IORT enables the unique opportunity of displacing organs at risk from the irradiation field by simple removal during the surgical intervention. Further on, the target area is defined and treated under visual control, which minimizes the risk of a geographic miss and no additional safety margins are needed to compensate for interfraction movements. Although the applicability of the linear-quadratic model for high single doses as used in IORT has been questioned to some extend , it can be assumed from clinical and experimental data that a combination of 15 Gy IORT boost with 45 Gy EBRT dose as used in our study is biologically equivalent to ≥70 Gy EBRT in conventional fractionation . The translation of dose escalation into improved local control found in our study has also been reported by other investigators using IORT in locally advanced unresectable pancreatic cancer: For example, Garton et al.  found 1- and 2-year local control rates of 86% and 68% and a median overall survival of 15 months in a series of 51 patients with unresectable pancreatic cancer treated with neoadjuvant EBRT followed by IORT. Roldan et al  compared patients with locally advanced pancreatic cancer scheduled either for IORT + EBRT or EBRT alone and observed a significant advantage in local control in favour of the IORT group. Mohiuddin et al.  found a local control rate of 69% with a median overall survival of 16 months in 49 patients with unresectable pancreatic cancer treated by IORT and postoperative chemoradiation followed by maintenance chemotherapy. Moreover, Willett et al.  reported a median survival of 13 months in a large series of 150 patients with unresectable pancreatic cancer treated with a combination of IORT and EBRT. These and our results compare favourable with approaches of chemoradiation or chemotherapy alone in terms of local control and strengthen the value of locally controlled disease for overall survival as supported by a recent pooled analysis on behalf of the International Society of Intraoperative Radiation Therapy (ISIORT) in Europe . This described a prolonged survival in resectable pancreatic cancer patients who remained free of local recurrence for more than two years after surgery and IORT (5-years OS 28%) compared to those who failed locally (5-year OS 0%).
However, beside the encouraging local control rate found in our series, distant failure occurred in the majority of patients. In contrast to primary pancreatic cancer, the value of adjuvant chemotherapy after resection of a local recurrence or the use of additional chemotherapy after chemoradiation has not been clearly established. Therefore, only a minority of our patients received additional chemotherapy directly after completion of local treatment at the discretion of the treating medical oncologist. Given the known advantages of adjuvant chemotherapy in terms of disease-free and overall survival after gross total resection of primary pancreatic cancer , additional adjuvant systemic therapy might be beneficial also after local treatment of an isolated local recurrence. This should be evaluated, although a recent retrospective analysis by Baschnagel et al.  failed to show a benefit from additional systemic therapy after postoperative chemoradiation in primary pancreatic cancer cases. The possible value of additional systemic therapy seems to be further supported by the fact, that those studies which report encouraging results with chemoradiation for locally advanced pancreatic cancer, additional or maintenance chemotherapy was used after completion of local therapy [13, 15].
Another issue in pancreatic cancer is treatment toxicity. In our study, we observed a 90-day postoperative mortality rate of 3% (1 patient). Severe postoperative complications were found in 8% of the patients. Acute gastrointestinal (GI) grade III toxicities attributable to chemoradiation were observed in 10% and grade III hematological toxicities in 20% of the evaluable patients. No severe IORT related toxicity was found. These figures compare favorable with data from the literature. Calvo et al.  reported 33% grade III GI-toxicity and 6% severe hematological toxicity with neoadjuvant chemoradiation + IORT. Aristu et al.  observed 12% severe gastrointestinal and 16% hematological toxicity during neoadjuvant chemoradiation followed by IORT in unresectable cases. IORT can be safely applied in patients with pancreatic cancer without an increase of postoperative morbidity or mortality compared to surgery alone as shown by Reni et al.  in a comparative series of 203 patients. Considering toxicity of external beam chemoradiation, it has to be emphasized, that severe toxicity is closely associated with the radiation technique and the irradiated volume. If modern radiation techniques are used, and the radiation fields are restricted to the gross tumor volume with small safety margins, the incidence of severe GI-toxicity appears to be much lower than in previous trials using elective nodal irradiation as well. A recent EORTC trial comparing chemotherapy with gemcitabine mono versus concurrent chemoradiation with gemcitabine in the adjuvant setting of pancreatic cancer observed no significant difference in toxicity between the treatment arms .