Molecular Profiling of Pancreatic Cancer Patients—Letter

We read with great interest the article by Pishvaian and colleagues (1) that explored the clinical implications of molecular profiling of pancreatic adenocarcinoma and reported various actionable genes that may provide guidance for clinicians to practice precision medicine. The authors reported improved outcomes in patients with pancreatic adenocarcinoma who were treated with "matched therapy" based on the molecular profiling compared with patients treated with "unmatched therapy." The authors also argued that 27% of those patients who underwent genetic profiling were found to have "highly actionable" alterations. Although, we agree that molecular profiling of pancreatic adenocarcinoma may uncover important molecular information that are "potentially" actionable, we would like to discuss a few points in regards to data presented in this study.

First, mutations in homologous recombination (HR) DNA repair pathways genes such as BRCA1/BRCA2, and PALB2 render sensitivity to DNA-damaging agents, particularly platinum-based therapies (2). There is also growing evidence that these mutations may also generate actionable genes as potential therapeutic targets for PARP inhibitors (3). However, it is unclear at this time if this therapeutic effect remains active after disease progression on platinum-based therapy. In a recently published phase II clinical trial of 16 patients with HR-deficient pancreatic adenocarcinoma (of those 14 received prior platinum-based therapy), no objective response was reported and best response was stable disease only in 4 patients (4) suggesting that PARP inhibitors as a single agent may have limited, if not absent, benefit in postplatinum phase. Therefore, classification of BRCA1/2-mutant patients with previous platinum exposure in "unmatched therapy" group is not evidence based and might have impacted reported survival outcomes in this study. Moreover, some of those "highly actionable" alterations have shown to be therapeutic targets in other cancers and there are limited data in pancreatic cancer. Molecular behavior of pancreatic adenocarcinoma is different and efficacy of targeting agents on these pathways may not yield same effect. For example, sonic hedgehog inhibitors have been shown to be effective in basal cell carcinoma (5); however, they did not show any significant activity in pancreatic adenocarcinoma (6) although many studies have identified increased expression of sonic hedgehog signaling in pancreatic cancer (7). Therefore, extrapolation from other cancers based on individual patient response may lead misinterpretations and inappropriate categorization of patients as matched versus unmatched therapy. Finally, it is also important to note that the presence of BRCA1/2 mutations has also shown to be prognostic in pancreatic cancer (8) beside their predictive value, and this point has not been addressed in survival outcomes analyses in the article by Pishvaian and colleagues (1).

Overall, there is strong evidence and value for performing molecular profiling of individual pancreatic adenocarcinoma to better tailor treatments for this challenging disease as well as assignment of patients for clinical trials; however, it is important to note that there are many limitations in this article beyond patient heterogeneity and nonrandomization to draw the conclusion of superior outcomes in "matched therapy" group, particularly due to extrapolations from published data in other cancers and lack of available evidence in categorization of patients as "matched therapy" versus "unmatched therapy" as briefly discussed above.

See the Response, p. 6612