FUT2 and FUT3-specific normalization of DUPAN-2 and carbohydrate antigen 19-9 in preoperative therapy for pancreatic cancer: multicentre retrospective study (GEMINI-PC-01)

Dear Editor Pancreatic cancer is one of the most challenging cancers to treat1. Some patients with unresectable disease at diagnosis may achieve a remarkable response by multimodal therapy and undergo subsequent surgery (so-called ‘conversion surgery’). However, determining the appropriate indications for conversion surgery often presents a formidable challenge1–3. In addition to carbohydrate antigen 19-9 (CA19-9), DUPAN-2, and FUT2 and FUT3, or FUT2/3 status (FUT2-null, FUT-intact, and FUT3-null) are emerging biomarkers for pancreatic cancer4,5. However, studies on integrating these biomarkers during preoperative therapy are lacking. The aim of this study was to examine changes in CA19-9 and DUPAN-2 levels according to FUT2/3 status during preoperative therapy and to test whether the determination of genotype-specific tumour marker normalization can enhance the usefulness of these markers when considering surgery. This study was the retrospective part of a multicentre observational study (GEMINI-PC-01/02). A total of 347 patients were enrolled who underwent pancreatectomy with curative intent after preoperative treatment for pancreatic cancer. After excluding two patients because of their final pathology and four patients because of mis-genotyping using a TaqMan-PCR panel (supplementary methods), which was designed considering the variant allele frequency in Japan (Table S1), data for 341 patients were used in the analyses. The patients’ characteristics stratified by resectability are shown in Table S2 (see Table S3 for metastatic disease). The distributions of CA19-9 and DUPAN-2 were more distinct for FUT2/3 status compared with resectability (Fig. 1a; see Fig. S1a for the sub-subset analysis). Overall survival did not differ greatly by resectability or FUT2/3 status (Fig. 1b and Fig. 1c respectively; see Fig. S1b for the sub-subset analysis). However, a significant prognostic difference across FUT2/3 status was observed in the unresectable subset (Fig. 1d). Therefore, the authors focused on the unresectable subset (Table S4) to develop a novel prognostic model, called the tumour marker gene model (TMGM). Using genotype-specific cut-offs, the authors categorized patients whose marker levels were lower than the cut-offs as TMGM-low and the remaining patients as TMGM-high (Fig. 1e; see Fig. S2 and the supplementary results for background information regarding the definitions). The TMGM successfully stratified patient prognosis compared with the compound single cut-off model (Fig. 1f; see Fig. S3 for the other models and Fig. S4 for the sub-subset analyses). The Cox proportional hazards analysis identified TMGM-high as an independent prognostic factor (Table 1). FUT2 and FUT3 status greatly affects carbohydrate antigen 19-9 and DUPAN-2 expression and might have impacted their prognoses, potentially due to the baseline changes of these markers. The Tumor Marker Gene Model was develped to demonstrated good prognostic predicting ability a CA19-9 and DUPAN-2 levels according to resectability (left-hand panels) and FUT2/3 status (right-hand panels). The y-axis is presented on a log-10 scale. b Overall survival curves according to resectability. c Overall survival curves according to FUT2/3 status. d Overall survival curves for the unresectable subset according to FUT2/3 status. Differences in survival were tested using the log rank or pairwise log rank test for post-hoc analysis. e Schema of the TMGM. The schema shows the distribution of CA19-9 and DUPAN-2 levels at surgery in the unresectable subset. The continuous lines indicate genotype-specific cut-offs and the broken lines indicate uniform upper limits (refer to the Supplementary material for the background information). An asterisk indicates that data were not available for the DUPAN-2 level. f Overall survival curves stratified by the compound single cut-off model (left-hand panel) and the TMGM (right-hand panel). CA19-9, carbohydrate antigen 19-9; R, resectable; BR, borderline resectable; UR, unresectable; F2N, FUT2-null; INT, FUT-intact; F3N, FUT3-null; TMGM, tumour marker gene model. Cox proportional hazards analysis for overall survival after surgery *New era chemotherapy includes gemcitabine plus nab-paclitaxel and fluorouracil, leucovorin, irinotecan, plus oxaliplatin (FOLFIRINOX). †Statistically significant. ‡CA19-9 >37 U/ml or DUPAN-2 >150 U/ml. RECIST, response evaluation criteria in solid tumours; TMGM, tumour marker gene model. Next, the TMGM was investigated further to verify its usefulness in practical applications. First, the TMGM and the DUPAN-2 single cut-off model were compared, which appears to be a good standalone predictor5. The TMGM was superior to the DUPAN-2 single cut-off model on the basis of Harrel’s concordance index (0.674 versus 0.569 respectively). Second, it was observed how the TMGM converged to TMGM-low, suggesting the best time to perform surgery. Data for 26 representative patients who had undergone preoperative therapy for 8–12 months were selected. The chronological changes in the two markers for each genotype showed more genotype-specific CA19-9 normalization compared with DUPAN-2 (Fig. S5). In summary, FUT2/3 status greatly affected CA19-9 and DUPAN-2 levels during preoperative therapy and the TMGM was successfully developed to stratify patient prognosis better than conventional single cut-off models. Unexpectedly, FUT3-null demonstrated a poor prognosis in the unresectable subset. Currently, the authors do not consider this finding a result of tumour biology but, instead, misinterpretation of tumour marker normalization, which is a pivotal surgical indication. Therefore, the authors decided to develop the TMGM with a focus on unresectable diseases. Physicians may have overestimated tumour marker normalization in this study because of deficient CA19-9 levels in FUT3-null patients. Some of them may benefit from avoiding surgery and continuing induction therapy. In contrast, potentially more FUT2-null patients could have undergone conversion surgery, even if they had relatively high CA19-9 and DUPAN-2 levels. Furthermore, conversion surgery for metastatic disease is beyond the scope of current guidelines. However, personalized treatment strategies can encompass conversion surgery as a possible option in select cases. The present study has certain limitations associated with its retrospective design. In particular, the study cohort included only patients who underwent surgery and the indications for surgery after preoperative therapy were not predefined. Second, the sample size of the unresectable subset was insufficient to consider the prognostic model robust. Prospective and/or large cohort studies, preferably registered at an early stage of induction therapy, are needed to validate the results of the present study. Haruyoshi Tanaka received a KAKENHI Grant-in-Aid for Young Scientists (Start-up) (JP21K20799), a KAKENHI Grant-in-Aid for Early-Career Scientists (JP22K15578), and grants from the Pancreas Research Foundation of Japan and the SOYU Medical Foundation, for the present work. The present study was pre-registered at https://jrct.niph.go.jp/latest-detail/jRCT1040230005, with the acronym GEMINI-PC-01/02 (jRCT1040230005). The authors thank Daigo Kobayashi, MD, Yoshikuni Inokawa, MD, PhD, and Yukihiro Yokoyama, MD, PhD, Department of Surgery, Nagoya University Graduate School of Medicine, Nagoya, Japan, for data acquisition. The authors also thank Akari Iwakoshi, MD, Department of Pathology, NHO Nagoya Medical Center, Nagoya, Japan, and Yosuke Tajika, MMSc, Department of Diagnostic Pathology, Faculty of Medicine, Academic Assembly, University of Toyama, Toyama, Japan, for material support. The authors thank Gabrielle White Wolf, PhD, and Jane Charbonneau, DVM, from Edanz (https://jp.edanz.com/ac), for editing the draft of this manuscript. H.T. and A.S. contributed equally to this study. Haruyoshi Tanaka (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Visualization, Writing—original draft), Ayano Sakai (Data curation, Formal analysis, Investigation, Methodology, Resources, Validation, Writing—original draft), Masaya Suenaga (Conceptualization, Data curation, Project administration, Resources, Writing—review & editing), Masamichi Hayashi (Data curation, Funding acquisition, Investigation, Project administration, Resources, Writing—review & editing), Tomohisa Otsu (Data curation, Investigation, Methodology, Project administration, Resources, Writing—review & editing), Nobuhiko Nakagawa (Data curation, Project administration, Resources, Writing—review & editing), Keisuke Kurimoto (Data curation, Project administration, Resources, Writing—review & editing), Mina Fukasawa (Data curation, Project administration, Resources, Writing—review & editing), Kazuto Shibuya (Data curation, Project administration, Resources, Writing—review & editing), Nobuyuki Watanabe (Data curation, Project administration, Resources, Writing—review & editing), Masaki Sunagawa (Data curation, Project administration, Resources, Writing—review & editing), Junpei Yamaguchi (Data curation, Project administration, Resources, Writing—review & editing), Takashi Mizuno (Data curation, Project administration, Resources, Writing—review & editing), Toshio Kokuryo (Data curation, Project administration, Resources, Supervision, Writing—review & editing), Hideki Takami (Conceptualization, Data curation, Project administration, Resources, Supervision, Writing—review & editing), Tomoki Ebata (Conceptualization, Project administration, Resources, Supervision, Writing—review & editing), Tsutomu Fujii (Conceptualization, Data curation, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing), and Yasuhiro Kodera (Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing) Haruyoshi Tanaka has received a one-time payment as part of a short-term advisory agreement with Minaris Medical Co., Ltd, Tokyo, Japan (currently Canon Medical Diagnostics Corporation, Tokyo, Japan). This relationship did not influence the present study and the payment was not used as a funding source for the research. A patent application related to the methodology described in this study is currently pending. This had no impact on the study’s outcomes or conclusions. The authors declare no other conflict of interest. Supplementary material is available at BJS online. The data generated in this study are available from the corresponding author upon reasonable request.