88 Discussion Our findings show that looking at TTT NOAs in all TTT parameters except for MLe at both feet should be considered as new approach to improve the consistency and balance between the selection of TTT parameters, measuring sites and definition of “abnormal QST”. We also demonstrate that the use of MLe leads to desensitization, further confirming that MLe should be discouraged as a diagnostic tool for SFN. Moreover, adding the SFNSL questionnaire could further improve diagnostic accuracy. To begin with, the PCA analysis showed that not all TTT parameters correlate with each other. Therefore, focusing only on the detection thresholds, as currently suggested, results in a loss of information. The fact that not all parameters correlate with each other, could be due to the fact that different specific functions are measured by different parameters. Additionally, clusters consisting of patients with probable SFN showed a large overlap with clusters consisting of participants without SFN. Consequently, no specific parameter was consistently abnormal in patients with probable SFN. It was not only within TTT measurements that a lack of correlation was found, but a lack of correlation between other diagnostic methods was also reported in the literature.21 We therefore assumed that expanding the number of measurements increases the reliability of SFN diagnosis. Our results confirmed that testing both feet did indeed represent a more accurate test side than the hands. To define a clinically applicable QST protocol, it was determined that measuring all TTT parameters and assessing the NOAs would essential. When CDT MLi, WDT MLi, TSL, PHS, CPT and HPT were measured at both feet using a cutoff of at least 2 abnormal parameters, the specificity of QST increased by 15% compared with the combination suggested in the new Besta criteria.7 Moreover, these measurements can be performed at both feet within 30 minutes, making them easy to apply. Combining TTT with a simple SFNSL questionnaire improved the diagnostic accuracy even further. Moreover, a higher TTT NOAs and higher SFNSL score was found in patients with sarcoidosis and probable SFN compared with sarcoidosis patients without SFN and healthy controls. Finally, a correlation between TTT NOAs and SFNSL score suggested a relation between the number of abnormalities in QST and the severity of SFN. Our study found four reasons for not measuring MLe. Firstly, we found a high degree of variance in the total number of measurements required to determine a detection threshold. This suggested inconsistent reactions by the participants, which could be explained with desensitization. MLe searched for the lowest level which can be detected, and needs confirmation with two times “yes” at the same lowest level. However, desensitization gradually increased the threshold, so the first “yes” at the lowest level changed to a “no” the second time. Consequently, more measurements were required to determine the detection threshold. This increasing number of measurements resulted in insensitivity of the skin to temperature stimuli, which could affect the measurements following the MLe measurements. The second reason was that the MLe parameter showed a lack of specificity, resulting in high percentages of abnormalities in healthy controls. This suggested that not only sarcoidosis patients, who often suffered from fatigue, were prone to desensitization, but that fit healthy controls also struggled with this complex method. Thirdly, since MLe should correct for time dependency, it was expected that MLe would result in normal findings in some cases where MLi resulted into abnormal findings. When we looked at the distribution of normal and abnormal results with MLi and MLe, MLe unexpectedly classified more participants as abnormal than MLi. CDT MLe measured 22% abnormal participants vs 17% with MLi and WDT MLe measured 11% vs 4% with MLi. A lack of correlation was also demonstrated by the low 5 92 5
RkJQdWJsaXNoZXIy MjY0ODMw