Abstract
IMPORTANCE: Quality assurance of molecular tumor boards (MTBs) is crucial in cancer genome medicine. OBJECTIVE: To evaluate the concordance of recommendations by MTBs and centrally developed consensus treatment recommendations at all 12 leading institutions for cancer genomic medicine in Japan using 50 simulated cases. DESIGN, SETTING, AND PARTICIPANTS: This was a prospective quality improvement study of 50 simulated cancer cases. Molecular tumor boards from 12 core hospitals independently recommended treatment for 50 cases blinded to the centrally developed consensus treatment recommendations. The study's central committee consisted of representatives from all 12 core hospitals in Japan who selected the 50 simulated cases from The Cancer Genome Atlas database, including frequently observed genomic alterations. The central committee recommended centrally developed consensus treatment. The concordance rate for genomically matched treatments between MTBs and centrally developed consensus treatment recommendations was evaluated. Data analysis was conducted from January 22 to March 3, 2021. EXPOSURES: Simulated cases of cancer. MAIN OUTCOMES AND MEASURES: The primary outcome was concordance, defined as the proportion of recommendations by MTBs concordant with centrally developed consensus treatment recommendations. A mixed-effects logistic regression model, adjusted for institutes as a random intercept, was applied. High evidence levels were defined as established biomarkers for which the treatment was ready for routine use in clinical practice, and low evidence levels were defined as biomarkers for genomically matched treatment that were under investigation. RESULTS: The Clinical Practice Guidance for Next-Generation Sequencing in Cancer Diagnosis and Treatment (edition 2.1) was used for evidence-level definition. The mean concordance between MTBs and centrally developed consensus treatment recommendations was 62% (95% CI, 57%-65%). Each MTB concordance varied from 48% to 86%. The concordance rate was higher in the subset of patients with colorectal cancer (100%; 95% CI, 94.0%-100%), ROS1 fusion (100%; 95% CI, 85.5%-100%), and high evidence level A/R (A: 88%; 95% CI, 81.8%-93.0%; R:100%; 95% CI, 92.6%-100%). Conversely, the concordance rate was lower in cases of cervical cancer (11%; 95% CI, 3.1%-26.1%), TP53 mutation (16%; 95% CI, 12.5%-19.9%), and low evidence level C/D/E (C: 30%; 95% CI, 24.7%-35.9%; D: 25%; 95% CI, 5.5%-57.2%; and E: 18%; 95% CI, 13.8%-23.0%). Multivariate analysis showed that evidence level (high [A/R] vs low [C/D/E]: odds ratio, 4.4; 95% CI, 1.8-10.8) and TP53 alteration (yes vs no: odds ratio, 0.06; 95% CI, 0.03-0.10) were significantly associated with concordance. CONCLUSIONS AND RELEVANCE: The findings of this study suggest that genomically matched treatment recommendations differ among MTBs, particularly in genomic alterations with low evidence levels wherein treatment is being investigated. Sharing information on matched therapy for low evidence levels may be needed to improve the quality of MTBs.