Milea Timbergen

319 clinical outcome 43-47 . In Chapter 5 , we used radiomics to create a prediction model for the differential diagnosis of DTF (DTF versus STS) and the different genetic mutations (S45F, T41A, and WT) observed in DTF. We showed that our radiomics model was capable of distinguishing DTF from STS based on T1-weighted and T2-weighted MR images, and the model outperformed the two experienced radiologists. Unfortunately, the model was not able to predict the CTNNB1 mutation type of the included DTF tumours. The results of our study encourages the use of computer-aided models in the diagnostic trajectories of rare diseases. After establishing the correct diagnosis, treatment decisions are based on the patients’ preferences, comorbidities, symptoms, tumor site and tumour size. Treatment decisions are preferably made in a multidisciplinary team meeting in the presence of a radiologist, a pathologist, a surgeon, an oncologist, and a radiotherapist, all with relevant expertise regarding soft tissue sarcomas and DTF. The increasing knowledge about the biology of DTF has taught us that active surveillance can be a safe approach 3, 48 . Consequently, the rate of surgery as an initial treatment therefore decreased from 55% to 42% between 2010 and 2015 in a French nationwide cohort study 49 . Three prospective clinical trials (NTR 4714, the Netherlands, NCT02547831, Italy, and NCT01801176, France) are conducted in Europe to evaluate the percentage of patients that do not need to receive an active form of treatment 50-52 . The results of those trials are still awaited. In an attempt to systematically evaluate the results of the active surveillance using retrospective data, we performed a systematic literature review in Chapter 6 . This study showed that a majority of patients undergoing active surveillance have either stable disease or partial response. The overall median percentage of patients with partial response in the included studies was 20% (95% CI 10-24) which demonstrates the capacity of DTF tumours of undergoing spontaneous regression. About one-third of patients shift to an active form of treatment. Although the active surveillance approach avoids the complications seen in active treatments, patients should be made aware of the benefits and risks to increase the chance of a successful active surveillance. Moreover, there is a need for identification of patients that will benefit from this approach. A way of stratifying the patients that will benefit from a certain treatment approach is to determine the CTNNB1 mutation type. Several studies implicate a prognostic role for these mutations specifically after surgical resection 29-32 . These studies imply that the S45F mutated desmoid tumour has an increased risk of recurrence compared to other CTNNB1 mutations and WT desmoid tumours. However, these studies were all small-sized 11