Previous recreational cold exposure does not alter endothelial function or sensory thermal thresholds in the hands or feet

What is the central question of this study? Does recreational cold exposure result in cold sensitivity and is this associated with endothelial dysfunction and impaired sensory thermal thresholds? What is the main finding and its importance? Previous cold exposure was correlated with cold sensitivity of the foot, which might indicate the development of a subclinical non‐freezing cold injury. Endothelial function and thermal detection were not impaired in cold‐sensitive individuals; therefore, further research is required to understand the pathophysiology of subclinical and clinical forms of non‐freezing cold injury.

in the 1940s (Kuht et al., 2019). The increased popularity of outdoor recreational activities has meant that the civilian population is also at risk. The 'dose' (magnitude and duration) of cold exposure required to induce NFCI is not known but is likely to vary between individuals (Burgess & Macfarlane, 2009) and also to be dependent on activity (Kuht et al., 2019). In addition, subclinical forms of NFCI may also exist in individuals frequently exposed to cold conditions for short durations during recreational activities such as windsurfing, surfing and open water swimming (Eglin, 2011;Eglin et al., 2017).
The pathophysiology of NFCI is poorly understood but is thought to include both vascular (Eglin et al., 2013) and neural dysfunction (Vale et al., 2017). However, to date, the existing literature has not included appropriate control groups who have been exposed to cold conditions but have not received a cold injury. In order to investigate the mechanisms underpinning NFCI, it is important to characterize the responses of individuals with varying previous exposure to cold to determine whether cold exposure per se can alter neural and vascular function.
Cold sensitivity is present in ∼70% of NFCI cases (Francis & Oakley, 1996). Even in the 'normal' uninjured population, some individuals may have a degree of cold sensitivity as a result of their recreational activities (e.g. windsurfers), although they have not been diagnosed with a NFCI (Eglin, 2011). The cold sensitivity may be a result of compromised vasodilatation, because glyceryl trinitrate, an endothelium-independent nitric oxide donor, was found to increase the rate of rewarming after a mild cold challenge in individuals with cold sensitivity (Hope, Eglin, Golden, & Tipton, 2014). In addition, individuals of African or Caribbean origin, who are more susceptible to NFCI than their Caucasian counterparts (Burgess & Macfarlane, 2009), have been shown to have a reduced vasodilatory response to ACh (Maley, House, Tipton, & Eglin, 2015). These studies indicate that the underlying mechanism of the cold sensitivity associated with NFCI is endothelial dysfunction.
Sensory thermal thresholds (STTs) are impaired in individuals with NFCI (Oakley & Lloyd, 1990;Vale et al., 2017). These changes in thermal sensation may be long lasting, if not permanent (Oakley & Lloyd, 1990). If individuals who demonstrate cold sensitivity in the absence of a cold injury diagnosis have a subclinical condition, it is postulated that they might also show reduced thermal sensitivity.
Therefore, in this study we investigated: (i) the effect of cold exposure experienced during recreational activities on peripheral vascular function and STT; and (ii) whether subclinical NFCI (cold sensitivity) was accompanied by endothelial dysfunction and impaired STT. Our first hypothesis was that prior cold exposure would be negatively correlated with peripheral vascular function (i.e. greater cold exposure would be associated with a lower skin blood flow response to transdermal delivery of ACh) and positively correlated with STT (i.e. greater cold exposure would be associated with higher STT, indicating poorer thermal sensitivity). Our second hypothesis was that cold-sensitive individuals would have impaired endothelial function and STT compared with age-and sex-matched control subjects.

New Findings
• What is the central question of this study?
Does recreational cold exposure result in cold sensitivity and is this associated with endothelial dysfunction and impaired sensory thermal thresholds?
• What is the main finding and its importance?
Previous cold exposure was correlated with cold sensitivity of the foot, which might indicate the development of a subclinical non-freezing cold injury.
Endothelial function and thermal detection were not impaired in cold-sensitive individuals; therefore, further research is required to understand the pathophysiology of subclinical and clinical forms of non-freezing cold injury.

Ethical approval
The protocol was approved by a local research ethics committee (SFEC 2016-031), and all volunteers gave informed, written consent before participation. The study conformed to standards set out in the Declaration of Helsinki (2013), except for registration in a database.

Participants
A total of 27 healthy volunteers (15 men and 12 women) with a range of previous cold exposure participated in the study ( Table 1).
The greatest cold exposure reported was by a frequent open water swimmer, who also completed an ice mile without a wetsuit. Intermediate cold exposure included participants who reported regularly undertaking short sea swims or dips or who undertook water sports, such as dingy sailing or kite surfing, throughout the year. Participants who reported frequently undertaking outdoor activities such as crosscountry running, mountain biking, football or rugby were considered minimally cold exposed. Participants who reported undertaking no regular outdoor activities in the last 2 years were considered non-cold Cold exposure ranking  18.2 (6.3) 11.6 (8.5) U = 22; P = 0.114 were formed, who were closely matched for age, sex, physical activity and anthropometry (Table 2; n = 9 in each group). In addition, they undertook the testing at the same time of day.
Participants undertook three tests. In all but two cases (owing to the availability of the participants to attend the laboratory), these were conducted on the same day in the following order: STT, endothelial function and CST. Testing was conducted in Portsmouth, UK between June and July 2016, when the mean outdoor temperature was 18.3 (1.8) • C.
On arrival at the laboratory, the height and mass of participants were measured using a stadiometer (Bodycare, Leicester, UK) and digital weighing scales (770, Seca, Hamburg, Germany), respectively.
Skinfold thickness was measured using skinfold callipers at the biceps, triceps, subscapular and suprailliac. Hand and foot volume were calculated using a water-displacement method by immersing the foot to the most prominent part of the external malleolus and the hand to the styloid process of the ulna. Sensitivity to touch on the toe pads and finger pads was assessed using the Ipswich touch test (Sharma, Kerry, Atkins, & Rayman, 2014). Female volunteers were also asked about their menstrual cycle to determine whether they were in the follicular or luteal phase or whether they were peri-or postmenopausal. However, the phase of the menstrual cycle was not controlled for because reproductive hormone status does not affect the responses to local cooling (Charkoudian, Stephens, Pirkle, Kosiba, & Johnson, 1999;Lunt & Tipton, 2014), thermal perception (Lunt & Tipton, 2014;Söderberg, Sundström Poromaa, Nyberg, Bäckström, & Nordh, 2006) or iontophoresis of ACh (Ketel et al., 2009).
Current physical activity was assessed using the International Physical Activity Questionnaire (Craig et al., 2003). Estimated physical activity was calculated by multiplying the reported duration of vigorous activity by eight, moderate activity by four and walking duration by 3.3 over the previous 7 day period to give MET-minutes per week (Craig et al., 2003). The DN4 questionnaire (Bouhassira et al., 2005) was used to identify whether participants had neuropathic pain.
Cold exposure was assessed with a cold exposure questionnaire used previously (Appendix 1; Eglin et al., 2017), which asked the participants to recall their previous cold exposure through school, work and leisure activities throughout their life (before 12, 12-18 and after 18 years of age). For each of these phases, the participants were asked where they lived (geographically) and whether they had participated in any sports or activities that took place in cold/wet conditions, giving details including the type of activity, when this occurred (e.g. June 2014present), frequency, duration and estimated water/air temperature.
In addition, they were asked how they rated their whole body and hands/feet to cope with the cold (worse than average/average/better than average) and whether they had experienced any symptoms (numbness, swelling, redness, tenderness or tingling) after being exposed to cold/wet conditions. Finally, they were asked whether they thought they had either Raynaud's phenomenon or NFCI and whether this had been diagnosed medically and, if undiagnosed, to describe their symptoms. Given the reliability of recalling cold exposure might vary between individuals, rankings of cold exposure were estimated by examining the participants' reports of their cold exposure history in the previous 2 years, taking into consideration the frequency, duration and severity (air temperature and water temperature) of their exposure. This ranking [from 1 (greatest cold exposure) to 27 (least cold exposure)] was initially completed by two researchers (C.M.E. and H.M.) independently, after which any discrepancies in ordering were settled by a third researcher (J.T.C.). Cold sensitivity was determined from the results of the CST (see section 2.5).

Sensory thermal threshold test
Warm and cold STTs of the hand and foot were assessed using a thermal sensitivity tester (Physitemp Instruments Ltd, Clifton, NJ, USA) at an environmental temperature of 23.6 (0.5) • C, as previously

Endothelial function; response to acetylcholine
After a 20 min acclimation period to the ambient conditions [23.6 (0.5) • C], ACh was delivered transdermally using iontophoresis to three sites in the following order: the volar aspect of the left forearm, the middle phalanx of the middle finger of the left hand and the dorsal aspect of the left foot, as described previously .
Acetylcholine (Sigma Chemicals, Aldrich) was diluted with sterile water for injection to achieve a concentration of 1% w/v. The iontophoresis protocol consisted of four pulses of 25 µA followed by one pulse of 50 µA, one of 100 µA, one of 150 µA and a final pulse of 200 µA applied for 20 s, with 60 s intervals between each pulse, during which no current was applied. After an interval of 5 min, the protocol was repeated on the next skin site.
Flux data from the laser Doppler and iontophoresis controller were recorded using a data-acquisition system and software (Powerlab and LabChart 7; AD Instruments, Colorado Springs, New Zealand).
The The skin temperature adjacent to the iontophoresis site was measured using a skin thermistor (Grant Instruments) and recorded on a data logger (Grant Instruments).

Cold sensitivity test
The CST used in the present study has been described in detail elsewhere (Eglin et  skin temperature is 2.7 and 8.7%, respectively .
Skin blood flow was measured using a laser Doppler probe (VP1T/7; Moor Instruments, Axminster, UK) placed on the big toe pads during foot immersion and on the pads of the thumbs during hand immersion.
Skin blood flow was calculated using minute averages and expressed as CVC (flux/mean arterial pressure).
Thermal sensation and comfort of the immersed foot/hand were measured using 20 cm visual analog scales (from 0, extremely cold to 20, extremely hot; and from 0" very comfortable to 20 extremely uncomfortable) and recorded before immersion, during immersion and every 2 min of the rewarming period. Pain sensation in the immersed foot/hand was recorded using a 0-10 pain scale (Ferreira-Valente, Pais-Ribeiro, & Jensen, 2011) at the same time points.
During each test, environmental conditions adjacent to the participant were measured using a wet bulb globe temperature meter (Grant Instruments) and recorded every 5 min.

Data analysis
The assumption of normal distribution of data was assessed using descriptive methods (skewness, outliers and distribution plots) and inferential statistics (Shapiro-Wilk test Average toe and finger skin temperature and minute averages of big toe and thumb skin CVC during the CST were compared between CS and control groups before, immediately after immersion and at 5 and 10 min of the rewarming period, using a mixed-model

ANOVA [group (two factors) × time (three factors)] followed by
Bonferonni-corrected multiple comparison tests. Thermal comfort and sensation were compared between groups at the following time points: before immersion, during immersion, immediately after immersion, at minute 2 of rewarming and the average response over minutes 4-10 of rewarming, using a mixed-model ANOVA.

Cold sensitivity test
Mean toe skin temperature was significantly lower in CS compared with control subjects (F 1,48 = 151.8, P < 0.001; Figure 1a CONTROL CS F I G U R E 3 Median (individual data points) for warm and cold temperature thresholds of the foot and fingers of control (n = 9) and cold-sensitive (CS; n = 9) groups. ** P < 0.01 Thermal sensation and thermal comfort were similar between groups before, during and after foot and hand immersion (Table 3).
During foot immersion, four control participants and two CS participants reported mild pain. Three of these control participants and both CS participants also reported mild pain during hand immersion.
Only one individual in the control group reported mild pain during the rewarming period (hand only).   Ratings for thermal sensation and thermal comfort are as follows: 0 = extremely cold/uncomfortable; 10 = neutral; 20 = extremely hot/comfortable.  Average maximum cutaneous vascular conductance (CVC) and area under the curve (AUC) are given for both groups.  (Table 4). No differences in the skin blood flow responses to ACh in the forearm, finger or foot were observed between groups (Table 4).

Correlations
A moderate correlation was found between cold exposure ranking and mean toe skin temperature after 5 and 10 min of rewarming (r = 0.4083, P = 0.0345 and r = 0.4189, P = 0.03, respectively; Figure 4).
No significant correlations were observed with any other measures taken during the cold sensitivity, STT or endothelial function tests.

DISCUSSION
This is the first study to examine systematically the effect of recreational cold exposure on vascular function and sensory thermal thresholds. A moderate correlation was observed between cold exposure rank and toe skin temperature during rewarming after foot immersion. No other significant correlations were identified with cold exposure ranking, and therefore our first hypothesis is accepted, in part. Contrary to our second hypothesis, we did not observe any physiologically meaningful differences in endothelium-dependent vasodilatation or detection of warm or cold stimuli between the CS and control groups.
The peripheral vascular responses and sensory thermal thresholds were examined in 27 individuals with a wide range of previous cold exposure. Although some of these participants were frequently exposed to very cold environments during their leisure activities (winter sea swimming), none of the participants had NFCI or neuropathic pain. As expected, individuals with greater exposure to cold showed a greater degree of cold sensitivity, having lower skin temperatures during the rewarming phase of the CST; however, this was only a moderate correlation ( Figure 4). Interestingly, the participant with the greatest cold exposure in the last 2 years rewarmed relatively quickly after the 2 min foot immersion, whereas other participants with apparently limited cold exposure rewarmed slowly ( Figure 4). This might, in part, be attributable to self-selection, with those who are more 'cold tolerant' being more likely to participate in recreational activities involving cold exposure.
The situational risk factors that predispose individuals to NFCI during cold exposure include feeling generally cold and having static duties (Kuht et al., 2019). In addition, repeated hand immersions into water at 8 • C result in an attenuation of the cold-induced vasodilatation response and lower skin temperature (Daanen, Koedam, & Cheung, 2012;Geurts, Sleivert, & Cheung, 2005;Mekjavic, Dobnikar, Kounalakis, Musizza, & Cheung, 2008). Although the cold-exposed participants in the present study reported cold extremities, they were all undertaking physical activity during their cold exposures and, in many cases, strenuous exercise. Exercise, particularly involving the whole body, during cold exposure might therefore protect peripheral vascular function. Indeed, both exercise training (Keramidas, Musizza, Kounalakis, & Mekjavic, 2010) and a slightly elevated body temperature are known to augment the cold-induced vasodilatation response (Daanen, Van de Linde, Romet, & Ducharme, 1997). In addition, although cold sensitivity is a common long-term symptom of NFCI (Francis & Oakley, 1996), the severity of the cold sensitivity is variable and not related to the severity of NFCI (Eglin et al., 2013).
Cold exposure ranking was not correlated with the responses to either ACh or STT. This indicates that previous cold exposure, which does not result in cold injury, does not compromise endotheliumdependent vasodilatation or alter the detection of warm or cool stimuli. However, there are limitations associated with the cold exposure ranking, which relied on accurate recall, because retrospective self-reporting is limited by recall bias and might not be well suited to address how behaviour changes over time and across contexts (Shiffman, Stone, & Hufford, 2008). In addition, our subjective judgement of the cold exposure of the participants might augment this bias. The ranking was based on the previous 2 years of cold exposure, because it was thought that this would provide the most accurate recall. However, significant cold exposure earlier in life might have altered vascular function regardless of the current level of cold exposure. Some patients with NFCI still have cold sensitivity many years after their initial injurious cold exposure (Francis & Oakley, 1996).
This study has confirmed the differing response of the feet and hands to a cold challenge (Figure 1; Cheung, 2015;Eglin et al., 2017;Norrbrand, Kölegård, Keramidas, Mekjavic, & Eiken, 2017) despite the fact they are both likely to be exposed to the same environment, especially during swimming. In addition, endotheliumdependent vasodilatation was greater in the fingers compared with the foot (Table 2), but STTs were similar between sites. The reduced endothelium-dependent vasodilatory capacity in the foot might underpin the increased susceptibility of the feet to NFCI (DeGroot et al., 2003;Golden et al., 2013;Kuht et al., 2019). However, dependency and wet and tight footwear are also likely to be factors Kuht et al., 2019).
Despite having a lower toe skin temperature, the CS group reported similar thermal sensation and thermal comfort to the control group throughout the CST (Table 3). In addition, endothelium-dependent vasodilatation was not compromised in the CS group (Table 4).
Contrary to our second hypothesis, CS subjects appeared to be more sensitive to detecting cold stimuli in their feet than their control counterparts ( Figure 3). However, the magnitude of this difference was small (0.2 • C), as was the effect size (0.23), and therefore we did not consider this significant in practical terms. In contrast, patients with NFCI have a reduced ability to detect warm and cold stimuli (Vale et al., 2017).
The low toe skin temperatures and slow rate of rewarming in CS individuals might be attributable to impaired vasodilatory capacity, because their response to the CST can be augmented by administration of glyceryl trinitrate (Hope et al., 2014) but not through dietary nitrate supplementation . In the present study, the response to transdermal delivery of ACh was similar between groups (