Due to covid, the recruitment plan was disrupted and recruitment is on hold for a undetermined period of time.

The study aims to compare different methods to assess thermal detection ability in diabetic patients, as a way to monitor and diagnose neurological complications of diabetes mellitus.

Diabetic polyneuropathy is a frequent complication of diabetes mellitus. The impairment of peripheral nerve fibre function can be very variable, predominantly affecting large-diameter fibres (subserving touch), small-diameter fibres (subserving thermonociception), or both.

Thermal detection threshold evaluation can be used to quantify the extent of function loss (hypoesthesia) and, to a lesser extent, gain (hyperesthesia) in patients with thermonociceptive impairments. They are important features of quantitative sensory testing (QST) protocols (Rolke, Baron, et al., 2006; Rolke, Magerl, et al., 2006) and are pivotal to the determination of sensory phenotypes (Baron et al., 2017; Raputova et al., 2017). Their role is particularly important in the diagnostic workup of neuropathies affecting small fibers (i.e., the subgroup of primary afferents responsible for thermonociception and autonomic functions) such as painful diabetic neuropathies (Terkelsen et al., 2017; Tesfaye et al., 2010).

Currently, clinical measurements of thermal detection thresholds are mainly performed using the method of limits (Fruhstorfer, Lindblom, & Schmidt, 1976), in which a continuous heating or cooling ramp (usually at a slow rate, 1°C/s in the case of the DFNS QST protocol (Rolke, Magerl, et al., 2006)) is applied to the skin of the patient who is instructed to press a button as soon as he/she feels a warm or cold sensation. The detection threshold is then considered to be the temperature reached at the moment the patient pressed the button. The method of limits has been known for a long time to be methodologically biased due to its reliance on the reaction time (Yarnitsky & Ochoa, 1991), which lead to an overestimation of the threshold value corresponding to the temperature change that occurred between detection and it's signalling by a motor response. This is problematic as reaction times are under the influence of decision and motor reaction response speeds which may be affected by factors irrelevant to the assessment of sensory discrimination, such as cognitive or motor impairments.

A methodologically sounder approach for threshold measurement is the method of levels or constant stimuli (Kingdom & Prins, 2010). A number of preselected stimulus intensities are presented a number of times in random order and the subject is asked whether he/she felt each stimulus. Unlike the method of limits, this approach is not biased by decision speed and motor function. Furthermore, this method enables the fitting of a psychometric function (probability of detection as a function of stimulus intensity) to the results, therefore moving thermal detection performance assessments from the outdated High Threshold Theory framework to that of the currently leading Signal Detection Theory (Kingdom & Prins, 2010). Whereas High Threshold Theory conceptualized detection as an ON/OFF process (below threshold, no detection occurs, above threshold detection always occurs), Signal Detection Theory sees detection as a probabilistic process (each stimulus intensity is associated with a probability of detection). This theoretical framework implies to redefine the threshold as the stimulus intensity for which detection probability equates 0.5. In addition to the threshold, the psychometric function is also defined by its slope, i.e. the rate at which detection probability changes around the value of the threshold. . Unfortunately, the method of levels has some important drawbacks. First, it is time consuming as it requires collecting responses to a large amount of stimuli (usually several hundreds) (Gescheider, 1997). Second, the range of stimulus intensities must be approximately centered on the actual threshold value and cover the transition range of detection probability.

To overcome these limitations, several adaptive procedures have been proposed. These procedures actively adjust the intensity of the presented stimuli depending on the previous responses of the subject (Kingdom & Prins, 2010). In the present study, we implemented for the first time the Psi method (a Bayesian adaptive algorithm proposed by Kontsevich and Tyler (1999)) to estimate the thresholds and slopes of the psychometric function for heat and cold detection. This algorithm associates each potential values of slope and threshold with a probability, updates this probability distribution based on the response recorded after each stimulus presentation (detected/not detected), and selects the next stimulus intensity so that the response to its presentation maximizes the entropy (i.e. the uncertainty around the values of slope and threshold) reduction.

In this study, we will test healthy controls with the conventional method of limit and the new psi method, in order to establish normative values for the new test.

Inclusion Criteria:

-

Exclusion Criteria:

• Alcohol beverage intake >3 units/day
• Habitual substance abuse
• History of chemotherapy
• Scar or dermatological condition at the site of stimulation (forearm and hands, leg and foot)
• History of neurological, psychiatric or metabolic disorder other than Diabetes Mellitus (screening will be performed with the patient)
• Currently taking drugs that could induce neuropathy (screening will be performed with the patient)
• For healthy controls: Suffering of chronic pain

SAP will be available on the OSF repository by January 2021. Anonymized data will be freely available (stored on an OSF repository) as supporting material for publication.

SAP will be available on the OSF repository by January 2021. Anonymized data will be freely available (stored on an OSF repository) as supporting material for publication.

No restriction