Is complete pathologic response in pancreatic cancer overestimated? A systematic review of prospective studies

Background: In literature, percentages of pathologic complete response (pCR) in patients presenting with resectable (RES), borderline resectable (BLR) or locally advanced (LA) pancreatic cancer (PaC) after neoadjuvant treatment (NADT) are variable, ranging 0–33%. Those data come mostly from retrospective reviews of single centres. The objective of this systematic review is to assess the incidence of pCR. Methods: Following the criteria of the PRISMA statement, a literature search was conducted looking for prospective papers focusing on neoadjuvant treatment in PaC. Retrospective papers, other than ductal carcinoma histologies and trials including metastatic patients, were excluded from the present review. Data extraction was carried out by 3 independent investigators. Meta-analysis was performed with ProMeta3 Software (Internovi, 2015). PROSPERO registry: CRD42018095641. Results: The literature search of Embase, Cochrane and Medline with the terms “neoadjuvant OR preoperative”, “pancreatic OR pancreas” and “cancer OR adenocarcinoma OR tumor” led to the identification of 3128 papers. We restricted the search to humans, last 10 years and English language articles resulting in 1158 eligible articles to review. Extended paper revision led to the inclusion of 27 papers. Complete pathologic response ranged 0–11.11%, at the meta-analysis 4% (95% CI 3–5%), in prospective studies 0–9.09% and in prospective databases 1.63–11.11%. Conclusions: Pathologic complete response in pancreatic cancer is actually infrequent: high-quality studies provide a more reliable picture of neoadjuvant effects, high rates of pCR are reported in selected retrospective studies but it is overestimated.