PDAC ranks the 14th cause of cancer but represents the 7th leading cause of cancer-related death,[3] thus reflecting its inherent aggressiveness. On top of this, PDAC incidence has been increasing in the last decades and it is projected to represent one of the main cancers in the upcoming years.[4] Patients with PDAC have an impaired outcome, and limited therapeutic options are currently available.[26] In this scenario, personalized medicine appears as a desirable option to enhance prognosis for PDAC patients, and the incorporation of ctDNA analyses in the routine therapeutic armamentarium arising as a potential complementary tool.
The utility of ctDNA analysis is sustained on the tumor dynamics.[13, 24] ctDNA is released in peripheral blood from tumor cells undergoing apoptosis, necrosis or even due to metabolic secretion. Advances in genome sequencing including targeted methods such as NGS have made analysis of ctDNA feasible for clinical use. Accumulating evidence supports this approach at different time points along disease progression, including early diagnosis, detection of minimal residual disease after surgery, molecular characterization for selection of targeted therapy, and monitoring treatment response. Although advances have been done in lung, breast and colon cancer, evidence is less advanced in PDAC and needs further translation to the clinics.[13, 14, 24]
More than 90% of PDAC present KRAS variants, the most frequently mutated oncogene in cancer.[27]Although not targetable for many years, novel approaches have been developed for mutant KRAS tumors in the last decade, thus making some mutated variants of this gene druggable.[27, 28] On the other hand, non-mutant KRAS PDAC shows a greater number of molecular alterations with potential value for precision medicine. For example, pathogenic germline variants in different genes, such as BRCA1, BRCA2 or PALB2, or somatic-level fusions in NRG1, rearranged RET or NTRK.[29–32]
Consistent with the literature, we found KRAS and TP53 as the most frequently mutated genes. While the identified type and amount of genetic alterations in tissue was consistent to what had been previously reported (87.5% of patients presented pathogenic alterations, mostly variants in KRAS (81.3%), we detected less and different alterations in plasma ctDNA. Percentages of patients showing pathogenic alterations varied according to different time-points: 61.5% in baseline plasma samples, 39.1% in first tumor assessment plasma and 80% in progression plasma, thus reflecting the dynamics of tumor size. Also, proportions of each mutation changed in time. These differences probably reflect tumor heterogeneity and clonal evolution. When comparing tissue and baseline plasma analyses, plasma analyses could not detect pathogenic alterations in a percentage of the patients, which did indeed present alterations in their tissue. Previous studies assessing concordance levels between tissue and ctDNA NGS in gastrointestinal cancers have reported varying results depending on the experimental settings.[19, 21, 33, 34] This phenomenon could be explained due to the small amounts of ctDNA present in blood at certain settings, including early stage and non-shedding tumors.[12, 35] Additionally, these results can also be expected given the differences between the employed sequencing panels; i.e. the plasma panel (Oncomine™ Pan-cancer Cell-Free Assay) does not cover some of the genetic regions sequenced by the tissue panel (Oncomine™ Comprehensive v3 Assay). In summary, our results suggest that in the non-metastasic setting of PDAC, tissue-biopsy should remain as the gold standard to characterize tumor genetics, while ctDNA analysis are still investigational. Possible methods to overcome its sensibility and sensitivity may consider the collection of larger amounts of plasma. Nevertheless, we could show how plasma ctDNA analysis play a complementary role to tissue-biopsy, by improving molecular characterization. We showed some pathogenic variants in TP53 detected in plasma but not in tissue samples. Moreover, baseline plasma analyses revealed pathogenic variants in cancer-related genes beyond KRAS and TP53 in patients with no tissue samples available. Indeed, ctDNA was suggested to be more informative in the metastatic setting, being able to recover the genetic heterogeneity derived from different lesions, some of them potentially druggable, therefore identifying candidates for treatment with targeted therapies.[34]
The detection of cancer at the early stage of disease has the potential to greatly improve patient outcomes and reduce cancer-related mortality.[36] There is evidence supporting its utility at diagnosis with methylation-based sequencing ctDNA assays.[37] This is important especially for PDAC, where the only curative option includes a R0 resection (> 1 mm between tumor cells and margins) for early stages.[1, 26] Our analysis did not include methylation-based tests, but focused on genotyping. We were able to demonstrate a poorer outcome in patients with pathogenic alterations in baseline plasma, compared to those with non-pathogenic alterations (i.e., patients presenting no genetic alterations or only VUS or likely pathogenic variants). When focusing on KRAS variants, patients did not differ in prognosis, although this was probably related to the low number of KRAS mutated patients.
Resectability of non-metastatic PDAC patients mainly relies on the tumor vascular invasion. In patients presenting borderline PDAC, [7, 8] induction chemotherapy treatment is recommended, with the aim of eradicating micrometastatic disease and down staging the tumor, and with the intention of attempting secondary curative surgery.[1, 26] Our analyses showed how all the patients with borderline PDAC with non-pathogenic alterations in baseline plasma reached disease control after NAT, thus being suitable for surgery. By contrast, only the 28.6% of the patients with pathogenic alterations reached disease control after NAT (p = 0.002). Pathogenic alterations reflect the tumor mutational load and can be used to monitor treatment response and to identify resistance mutations at progression.[38–40]
Type of neoadjuvant chemotherapy is still a matter of debate.[9, 10] Considering the purpose of this approach, identifying genetic alterations to guide treatment decisions would be crucial. Preference of platinum combinations in patients with mutations in the homologous recombination genes[11] or targeted therapies in patients with druggable mutations would enhance treatment results. We could observe some pathogenic variants in different cancer-associated genes including BRAF, FGFR2 or ERBB3 that, although not currently available in the clinical practice, may be considered for treatment in future clinical trials.
Finally, when considering the results on the ctDNA analyses performed after the neoadjuvant treatment and prior to surgery (at the first tumor assessment), patients with non-pathogenic alterations tend to present higher rates of disease control than patients with pathogenic alterations. This phenomenon could somewhat be explained due to the effects of chemotherapy, stimulating tumor shrinkage and therefore minimizing tumor mutational load.[41]
Higher VAF of pathogenic genetic variants in circulating tumor DNA is an indicator of poor outcome.[38] We could observe how patients with baseline plasma samples showing higher VAF of pathogenic variants presented a worse prognosis than the others. A tendency was shown when considering samples taken at the end of the neoadjuvant treatment and prior to surgery. Similarly, the VAF sum was greater at the progression plasma samples, reflecting the progression of the disease, in agreement with previous studies, suggesting that ctDNA detection associates with tumor mutational load.[17, 35, 42] Specifically, we were able to demonstrate the utility of VAF dynamics in two patients, in which KRAS and TP53 variants from plasma samples were longitudinally followed. They were detected in baseline plasma, disappeared after neoadjuvant treatment prior to surgery, and were again detected at a higher VAF in the sample taken at progression.
Limitations of our study include the small number of patients enrolled., Additionally, the type of sequencing panels used, Oncomine™ Pan-cancer Cell-Free Assay for plasma samples analysis and Oncomine™ Comprehensive v3 Assay for tumor tissue. Oncomine™ Comprehensive covers 161 cancer-related genes, while Oncomine™ Pan-cancer is designed to detect alterations in 52 genes, thus pathogenic genetic alterations in other genes cannot be ruled out. Furthermore, post-surgical samples were missed, thus impeding the evaluation of minimal residual disease in this setting.