Consistency of a clinical decision support system with molecular tumour board recommendations for tumour sequencing-guided treatment of pancreatic cancer

Background

Pancreatic ductal adenocarcinoma (PDACs) can have actionable genomic alterations at varying frequencies. Harnessing tumour profiling for its treatment is an intricate procedure because the matching of aberrations with therapeutic opportunities is increasingly complex. We evaluated a clinical decision support system (CDSS) that can be used to rationalise this process.

Methods

Patients with advanced PDAC were enrolled in a prospective observational study to assess a CDSS, MH Guide. Sequencing data were analysed with the CDSS, and a study-specific molecular tumour board (MTB) assessed reported actionabilities, ineffective drugs, and excess toxicity warnings. We carried out a post hoc analysis of each patient’s treatment within the purview of their medical team and compared it with CDSS and MTB statements.

Results

We included 39 patients in the study, 31 had complete CDSS reports and 28 were discussed at the MTB. The CDSS made 80 treatment suggestions based on 61 actionable variants. It highlighted 14 inefficacy marker–drug pairs based on 7 individual markers and flagged a total of 15 individual markers of increased risks of drug-associated toxicity for 28 different cancer treatments. The study-specific MTB endorsed molecularly informed therapeutic options in 21 cases, but there was no exclusion of any drugs based on inefficacy or toxicity markers. The overall evidence underpinning assertions of actionability was weak. After the end of the study, eight patients received targeted treatment without signs of response or clinical benefit.

Conclusion

The oncology CDSS MH Guide can be deployed along the care path of patients with advanced PDAC but a critical review of assertions from it is warranted.