508 chapter 13 Challenges in research In the following paragraphs, the current and future challenges in research will be discussed, as arisen from our personal experience from the EfFECTS trial. This begins with challenges in thyroid research, such as the diagnosis of oncocytic thyroid nodules, pathophysiology of [18F]FDG-positivity in (benign) thyroid nodules, and active surveillance as a reference standard, followed by the more general contemplations in [18F]FDG-PET/CT research. Next, the challenges in diagnostic trials will be discussed, such as time limitations and more specific methodological considerations for diagnostic randomized-controlled trials, such as selection bias, blinding, diagnostic misconception and misestimation, diagnostic confidence and treatment compliance. Finally, I will elaborate on the challenges of cost-utility studies and open data. Challenges in thyroid research Oncocytic thyroid nodules As elaborated in more detail in our systematic review (Chapter 2), thyroid nodules with oncocytic cytology (i.e., Bethesda IV follicular neoplasm, oncocytic type) have been Achilles’ heel to many diagnostic techniques for years, with several studies concluding that their index test was accurate in all except oncocytic nodules [25, 170, 296, 309, 351]. The generally assumed more aggressive nature and less favourable prognosis of oncocytic thyroid carcinomas as compared to their nononcocytic follicular counterparts emphasizes the need for an accurate diagnostic workup [503, 504]. The aggregated findings from the studies in this thesis endorse that the accurate diagnosis of oncocytic thyroid nodules prior to diagnostic surgery remains challenging to date. But above all, these results exposed how little we may still know about the genomics, biological behaviour, and accurate preoperative diagnosis of oncocytic nodules as compared to non-oncocytic nodules. We have only in recent years come to realize that oncocytic nodules should always be assessed separately from nononcocytic nodules, for example, using separate imaging thresholds or specific molecular panels [25, 59, 480]. In Chapter 4, we showed that nearly all oncocytic nodules are visually [18F]FDG positive. This confirmed the suspicions of earlier studies that visual assessment of [18F]FDG-PET/CT does not differentiate between benign and malignant oncocytic nodules and should not be performed [59, 501]. Our SUV-based assessments of the [18F]FDG-PET/CT in Chapter 5, however, suggested that the SUVmax and other SUV derivatives may accurately classify oncocytic nodules, provided that separate SUV thresholds are applied to nononcocytic and oncocytic nodules [455]. The threshold SUVmax of 5.2 g/mL, at which 100% sensitivity and 100% NPV were observed in our study cohort, is in
RkJQdWJsaXNoZXIy MTk4NDMw