506 chapter 13 [695, 725]. Some of the molecular alterations that can be detected by these panels are associated with a low (<30%) rate of malignancy and are considered test negative, though they are formally defined as a (likely) pathogenic alteration (i.e., IARC class 4 or 5) [687, 698, 725]. While the binary categorization of MD results in ‘positive’ and ‘negative’ benefits the calculation of diagnostic accuracy parameters such as sensitivity and specificity, these low-malignancy-rate test-negative alterations may also threaten the NPV [687, 698, 725]. Nodule sampling error seems like a reasonable possibility when performing MD on FNAC specimens. Although one could logically assume that the risk of sampling error and false-negative results of cytology and MD would apply more to larger nodules (i.e., larger than 4 cm), this has not been unequivocally demonstrated in literature [698, 741, 746, 747]. Even though NGS only requires a very small amount (5 to 10 ng) of nucleic acids, nondiagnostic results due to insufficient quality and/or quantity of the cytology specimens are regularly described and are more likely when cytology smears are used instead of FNAC samples collected in preservative solution [118, 160, 704-706, 725]. Nondiagnostic results are considered a relevant downside of MD, as repeating the FNAC would subsequently be required to obtain new material for repeatMD procedure, either resulting in additional patient burden and health care-associated costs. As described in Chapter 12, we observed 13% overall nondiagnostic results in our MD analyses, with a maximum of 45% (44 of 97) failed fusion analysis (i.e., partially failed MD). Fusion analysis requires sufficient quality of RNA, which is more unstable than DNA due to its single-stranded structure and ribose sugars instead of deoxyribose sugars [748]. As such, it degrades faster during smear preparation and fixation, making it more difficult to preserve [705]. While we hypothesized that the age of cytology smears may have influenced the nondiagnostic rate, we found no statistically significant difference between this and the age of the specimens in our cohort [725]. However, the most recently tested smears were already approximately 3 years old. We hypothesize that the MD nondiagnostic rate may have been lower if fresher cytology smears could have been used for our analysis. MD-based diagnosis and treatment: a pivot shift? Diagnostic accuracy is defined by comparing the results of the index test with those of the reference standard in the same patients. The reference standard is defined as the best method for establishing the presence or absence of the index disease [30]. But what if the reference standard changes? And what if it changes by increasing knowledge generated by the index test? With the growing knowledge of thyroid genomics and concurrent molecular alterations, our overall point of view regarding the diagnosis of thyroid nodules is shifting. With growing accessibility to these advancing techniques, MD-based classification systems are without a doubt the future of
RkJQdWJsaXNoZXIy MTk4NDMw