Although Response Evaluation Criteria in Solid Tumors (RECIST) criteria [12] have been well established in other solid tumors, as well as, in metastatic PDAC, its usefulness in response assessment in BRPC/LAPC seems questionable (Table 9.1). In a meta-analysis of 111 trials including more 4394 patient undergoing neoadjuvant treatment for PDAC, only 6 studies explicitly used RECIST criteria for response evaluation and less than 40 % of trials clearly stated their criteria to assess tumor response [13]. Similarly, in trials specifically evaluating neoadjuvant therapy in BRPC, RECIST criteria are used in less than 50 % of cases [14]. In the above-mentioned meta-analysis, the CR rate of all trials (i.e., including those with unclear definitions) was just 3.8 % (95 %CI: 3–4.9 %) and the PR rate 29 % (95 %CI: 26–34 %). Katz et al. could show that in 122 well-characterized BRPC patients at a single-institution, complete remission (CR) following neoadjuvant therapy as defined by RECIST did not occur at all, and that partial response (PR) was rare, occurring in only 15 patients (12 %) [15]. Similar results have been reported by other groups [16, 17].

Some studies have used the NCCN definitions of resectable, BRPC, LAPC, or metastatic disease to describe response to preoperative therapy. Similar to RECIST data, true radiologic downstaging seems to be a rare event and occurred in just 1 % of the aforementioned study by Katz et al. [15]. In contrast, disease progression and upstaging occurs in roughly 20 % of patients independent of the primary tumor classification [13, 15, 16]. Preliminary data indicate that tumor downstaging might be increased using novel regimes, including FOLFIRINOX and nab-Paclitaxel, which have shown high response rates in the metastatic setting [18–20]. However, radiologic imaging still does not correlate with surgical resection or pathologic response parameters [20] (Fig. 9.1).

Other unbiased measures of radiologic response evaluation like radiographic tumor stage following treatment (T-stage) (Table 9.1) or primary tumor diameter before and after preoperative therapy have rarely been reported [14]. Similar to the results of RECIST criteria, the objective response rates for these outcome metrics have been low in BRPC/LAPC [13, 16].

Very limited data exists on positron emission tomography (PET) imaging for response assessment and it has not been tested in large prospective trials so far [21].

As one of the main arguments in favor of preoperative therapy is the “sterilization” of the tumor bead and occult micrometastases the pattern of recurrence might be used as an important response criterion, but has as yet not been widely used [22]. The ability of PDAC to spread along nerve routes and ganglia causing high rates of local recurrence [23] might be reduced using preoperative regimes. Therefore, tight and standardized radiologic follow-up is recommended in clinical trials of preoperative therapies (Fig. 9.1).

Importantly, all radiologic response measures do not adequately predict surgical exploration and more important resection rates (see below) [13, 16, 24, 25] (Fig. 9.1). While diagnostic sensitivity for arterial involvement as one of the main features of LAPC has been described as high as 97 %, sensitivity has been poor with rates between 67 and 91 % [26–29]. The reason for the inadequate radiological staging of neoadjuvant-treated PDAC seems to be the extensive peritumoral desmoplastic and inflammatory reaction elicited by pancreatic tumor growth and the surrounding stroma [30]. Consequently, microscopic complete pathologic resections (R0) are more frequent than would have been expected by imaging and intraoperative findings (see below).

Until results from ongoing trials with high methodological quality, clear BRPC and LAPC definitions, and sound response classifications based on objective criteria are available [31], the clinical relevance of radiologic response evaluation is limited. Based on current data, radiologic imaging cannot be used to predict successful (R0) surgical resection and its clinical use is therefore limited [15, 17].

Resection rates have widely been reported in studies investigating preoperative treatment of BRPC and LACP. In a recent meta-analysis, resection rate was 50.7 % (95 %CI: 44.0–57.4 %) in all patients (i.e., resectable and unresectable at primary diagnoses) undergoing neoadjuvant therapy. As expected resection rates were higher in patients deemed initially resectable (73.6 %; 95 %CI: 65.9–80.6 %) than in patients deemed unresectable at primary diagnosis (33.2 %; 95 %CI: 25.8–41.1 %). Similar rates have been reported in other meta-analysis focusing on BRPC patients [16], radiotherapy [32, 33], phase II trials [34], or specific chemotherapy regimes [35].

These results are frequently interpreted as arguments in favor of neoadjuvant therapy, but have to be interpreted with caution. Surgical resection rate is a highly biased endpoint for preoperative response assessment as: (a) most studies lack clear definitions of resectable, unresectable, BRCP, and LAPC, i.e., inclusion criteria are unclear; (b) the current absence of adequate radiologic staging tools to differentiate between neoplastic and non-neoplastic desmoplasia leads to non-validated treatment decisions; (c) clear indications for surgical exploration and resection are lacking. Therefore, the question of whether these results truly reflect downstaging due to neoadjuvant treatment or whether similar resection rates could have been achieved with direct surgery remains unclear. Results from a multicenter database study indicate that for BRPC with venous involvement only, neoadjuvant therapy does not increase resection rate [36] and is consequently discouraged in some guidelines [10]. Furthermore, as pointed out above, resection rates are generally higher than would have been expected from radiological response assessment. Hence, indication for surgical exploration should be handled liberally until better and more reliable imaging criteria are available.