The present study tested the hypothesis that despite R0 status of resected PDACs, detectable residual disease, predictive of disease relapse, might be evidenced in the resected tissue by molecular identification of mutant KRAS. We examined the mutant KRAS status in tissue samples of patients with the best prognosis, i.e. the margins of R0 resected tissues of patients who underwent UFS and of good responder patients who had surgery after NAT.
Even if KRAS is mutated in > 90% of PDACs, the use of KRAS mutation status as a prognostic marker largely depends on the capacity of the technique to detect KRAS mutant variants. Indeed, they occur preferentially in the G12 position (98%) while G13 and Q61 positions are around 1% each (13). The G12 aminoacid substitutions arise from 6 variants. One hundred percent of the DNAs from resected tissues in which tumor remnants were expected (i.e. tumors for the UFS group and tumors of the ypT1 NAT sub-group) showed detectable KRAS mutations. Of note, 50% (3/6) of the ypT0 NAT sub-group displayed KRAS mutations, despite the fact that the primary tumor were supposedly cleared. Overall, we did not detect any KRAS mutation in only 1 ypT0 patient (3%, undetectable MAF in tumor tissue remnant and in margins). This observation is agreement with KRAS mutation frequency in PDAC (90–95% according to the studies, reviewed in (21)) combined with the capacity of our technique to detect 98% of the mutations (All G12 and G13 codons substitutions). Altogether, these observations suggest that even if tumor clearance is pathologically confirmed, KRAS mutations persist despite the NAT and the surgery. Although poorly described, micrometastases are supposed to exist, as early as the onset of the disease. Indeed, PDAC can produce circulating tumor cells (CTCs) with metastatic potential during the formation of the primary tumor, before malignancy can be detected by histological methods (22).
Patients with KRAS alterations have decreased OS, and specifically, the KRAS G12D mutation confers a worse prognosis in comparison with other KRAS alterations (21, 23). It would be interesting to identify KRAS mutations in our cohorts and evaluate the link with prognosis.
The main limit of our study is the small size of the cohorts. However, they appear representative of bigger published comparable cohorts, in terms of RFS and OS. Indeed, in agreement with published resected cohorts comparable to our UFS group (for example PMID: 29113659, with an OS of 22.1 months and RFS 11.0 months against 21.2 months and 13.9 months, our study). The NAT group selected good responders (complete pathological response, ypT0/1/N0). Of all the patients examined during the period, they represented 5.6%, which is in the same ranged as previously observed (24). This strategy yielded good performance (8 out 18, i.e. 44% of relapses, and a RFS and OS of 58.9 and 64.4 months, respectively), as expected by published results (25). In addition, our cohort yielded 17.6% of ypT0/1/N0 after chemoradiotherapy, which is in agreement with published data (25). Finally, as expected, RFS and OS in the UFS group were lower than in the NAT group. However, the NAT group selected patients with good response (ypT0/1/N0) to treatment, which likely artificially enhanced the difference in the prognosis as compared to the UFS group that was recruited prospectively. We also limited our analysis to the good responders of NAT, and it would be of interest to assess the margins of the ypT3/4/N0/N+.
Residual disease evaluated by KRAS mutation positivity in margins was previously evaluated. In the venous margins of 22 UFS patients, Turrini et al. detected KRAS mutations in 55% of histologically negative venous margins (15), which is similar to our finding here of 41% (7 out of 17) and to Kim et al. (14). However, by contrast to the cited studies, our survival analysis did not stratify patients according to their margin positivity. We detailed that the characteristics of the included patients. Adjuvant treatment cannot explain survival discrepancies. Interestingly, Kim’s retrospective cohort counted only 11% of T3 tumors for 89% of T1/T2 and 57% of N+. In the same way, Turrini’s patients were mainly T1/T2, 72.8% and 59% N+ and were recruited retrospectively. Our UFS prospective cohort included 88% of T3 tumors and only 12% of T1/T2 tumors, for 100% N+. Thus, it seems that Kim and Turrini’s cohorts are not comparable to ours, at least in terms of tumor grade, which might explain the different prognosis value of KRAS mutation presence in margins, which may be of clinical value for lower grade PDACs. More recently, Kim et al. found that patients showing a KRAS MAF > 4.19% had shorter DFS and OS in resection margins, regardless the R status. Indeed, they included R1 margins (43%), and it is not clear whether the 4.19% threshold was distinct from the positivity threshold (16), that we clearly defined in the present study. It is possible that this threshold identifies the R1 margins, hence its prognostic value. We rarely detected MAFs > 4.19% (2 out of 70 determinations), probably because we did not include R1 margins. Their patients had 73% of T1/T2 tumors and 73% of N+. It would be interesting to analyze their 27 patients with higher tumor grades.