J.Douglas Steele, John F. Dillon, John N. Plevris, Joao L. Hauer, Ian A. D. Bouchier and Peter C. Hayes
University Department of Medicine The Royal Infirmary, Lauriston Place. Edinburgh. UK (Received 14 January 1993)
Previous studies have shown that changes in the peripheral circulation occur in patients with liver disease. We have investigated the temperature of the hands of 26 patients and 13 controls using a liquid crystal contact thermography system. Significant differences in resting hand temperature and appearance were observed when the patients were categorized according to Child's score. The appearance depended upon the severity of liver disease; Child's A patients have warm hands of normal thermographic appearance; Child's B patients have cold hands with an abnormal thermographic appearance; whilst Child's C patients have warm hands with an abnormal thermographic appearance. After right-hand ice-cold water immersion, right-hand recovery was abnormally slow in Child's B & C groups compared with Child's A patients and controls. In addition, there was left-hand reflex vasoconstriction during the right-hand warm-up period in those with severe liver disease. With regard to autonomic function, no significant difference in right-hand temperature, Thermographic appearance or warm-up rate was detected between those with and those without autonomic neuropathy. These results indicate that autonomic nervous system dysfunction is not the predominant cause of these changes.
Severe liver disease is associated with a hyperdynamic circulation. This is characterized by increased cardiac output and heart rate, and decreased peripheral vascular resistance (1,2). Few studies have investigated the peripheral circulatory changes, and the underlying mechanisms have not been clearly identified. An alteration of plasma concentration of vasoactive mediators in particular those synthesized locally or those metabolized by the liver, may cause the haemodynamic disturbances, as may autonomic nervous system (ANS) dysfunction (3,4).
Liver disease may be associated with unusually warm (5) or cold hands (6,7), indicating peripheral circulatory abnormalities. Liquid crystal contact thermography detects the heat emission from the body surface and predominantly measures thermoregulatory anastomotic (arteriovenous shunt) skin blood flow (8). We have previously shown using this system, that abnormalities in the thermographic appearance of the hand can be detected in alcoholic cirrhosis. The object of this study was to use liquid crystal contact thermography to investigate whether there are changes in hand-skin blood flow in cirrhosis irrespective of etiology and to observe the effect of severity of cirrhosis. In addition, a possible correlation between altered blood flow and ANS dysfunction was investigated.
Material and Methods
Patients
Twenty-six patients with liver disease and 13 control subjects without liver disease were studied. None of the patients or controls was receiving any drugs or had any additional disorder known to affect the peripheral circulation (e.g. diabetes mellitus, thyroid disease, cardiac failure or Raynaud's syndrome). None of the subjects had consumed alcohol within 12 h prior to the test.
Thermography
We used a liquid crystal contact thermography system (Novamedix, Whitchurch, Hampshire, UK) with a temperature resolution of 0.8"C. A change in temperature in excess of 0.8"C resulted in a change in crystal color according to the manufacturer-defined scale. Three liquid crystal screens were used, each having a maximum temperature range of 4"C and a nominal, center of range temperature sensitivity of 28, 30, 32 and 34"C. Screen temperature measurements were found to involve negligible error when the measurements were investigated using an aluminum block heated to a known temperature.
The procedure adopted to obtain and analyze the data from the subjects was similar to that described in previous studies (5,9). Briefly, the procedure was carried out in a specifically allocated room with a stable temperature of 22"C and which was draught proofed. The patients were seated at a desk, with the thermography system in front of them, and were equilibrated with the ambient temperature for 20 min before testing. All measurements were performed by the same observer. A simple digital skin thermometer measured the approximate hand temperature of each subject to select the appropriate screen. The subjects placed both their hands on the screen for not less than 20 s, producing a stable thermal image, and then a color Polaroid photograph was taken as a permanent record. This was defined as the baseline temperature distribution. The subject then immersed the right hand in ice-cold water at 70C for 30 s. Following a pilot study, the temperature of 7"C was selected for the following reasons: patients could comfortably keep their hands in the water, we demonstrated good thermographic responses to this temperature and the temperature was easily reproduced by the addition of tap water to ice. A sequential set of photographs was taken of both hands immediately after removal and after 2, 5, 8 and 10 min.
Cardiovascular autonomic function tests
We used a standard battery of cardiovascular autonomic function tests to assess all the patients with cirrhosis, the heart rate responses to Valsalva manoeuvre, standing up, and deep breathing; and the blood pressure responses to standing up and sustained handgrip, as has previously been described in detail (10).
The results of the five cardiovascular reflex tests were compared with published normal ranges (10) and scored as normal, border line or abnormal. The results of the five individual tests were categorized to give a level of autonomic involvement (10).
Analysis
The photographs were analyzed for average hand temperature, thermographic appearance, percentage recovery of the right-hand thermal image, and reflex vasoconstiction and vasodilatation of the left hand, blindly by a single observer. The variation of the analysis was less than 4%.
A transparent plastic grid with 5 by 5 mm sq was placed over each photograph and the predominant temperature registered in each square was noted. The average hand temperature was calculated by dividing the temperatures by the sum of the squares in each photograph.
The thermographic appearance was defined according to the following method. The temperature of each finger was compared with the temperature of the adjacent finger-wide region of palm and the following scores awarded 0 if the finger was clearly cooler, 1 if there was no difference and 2 if the finger was clearly warmer. These scores were then summed for each hand, giving an overall thermographic appearance index ranging from 0 (completely anisothermal; fingers relatively cool), through 5 isothermal) to 10 (completely anisothermal; fingers relatively warm).
The percentage recovery of the right-hand thermal image was calculated by first determining the total number of squares visible on the thermographic image. The visible squares corresponded to hand regions exceeding an arbitrary threshold of 2.3 +/- 1.0"C below average and temperature at baseline; i.e. the lowest temperature which could be detected on any given screen. The percentage recovery of the right hand was then calculated by dividing the total number or squares visible on the thermographic image by the total number of squares visible at baseline. The area under the percentage recovery curve was calculated using a trapezoidal method to provide a single percentage recovery of the right hand index.
Left hand reflex vasoconstriction and vascodilatation corresponded to clear temperature changes of around 1"C. However, this change was close to the screen resolution of 0.8"C; therefore, each photograph was assigned to one of only three left hand reflex categories relative to baseline; hyperthermic, isothermic or hypothermic.
The severity of liver disease was categorized according Child's score (11) and the mean values of hand temperature, thermographic appearance and percentage recovery of the right hand indices were compared for each of the patient groups and controls using Students t-test. The numbers of patients in each left hand reflex category at a given time after immersion for each of the patient groups were compared with the corresponding values for the control group using the chi-square test. The results are given as mean +/- standard deviation and the p-values correspond to a two-tailed test of significance. Patients were also categorized according to the results of the ANS function tests (10) and the mean values of average hand temperature, thermographic appearance and percentage recovery of the right hand indices for each group were compared using the same method as described for the Child's classification.
Results
The mean hand temperature and thermographic appearance scores for each group are shown in Fig. 1. The mean hand temperature of the controls (31.3 +/- 2.60C) was significantly different (p<0.03) from the Child's B patient group value (28.2 +/- 2.9"C). The control group hand temperature was not significantly different from the Child's A and Child's C patient group temperatures and the difference between the mean temperatures of the Child's A and C groups was not significant.
The mean thermographic appearance score for the controls (3.85+/-1.91) was significantly different from the Child's B group value (1.44 +/- 1.74) (p<0.007) and Child's C (1.57 +/- 0.53) (p<0.002). The control group score was not significantly different. from the Child's A group and the scores for the Child's B and C groups were not significantly different. The fingers were never observed to be warmer than the palms.
In the case of the Child's B group, the significantly cooler value of mean hand temperature was attributed to the significantly cooler fingers (relative to the palm). However, in the Child's C group, the warm value of mean hand temperature together with significantly cooler fingers (relative to the palm) indicates that the palm temperature is abnormally warm.