The first study compared thermal hyperhemia in patients with systemic sclerosis with that in patients with primary Raynaud's phenomenon and healthy controls. We studied 60 consecutive subjects at the Inserm Clinical Research Center (Grenoble University Hospital, Grenoble, France) from January 2004 to October 2004: 20 patients suffering from systemic sclerosis, 20 patients with primary Raynaud's phenomenon and 20 healthy volunteers. These subjects are involved in a larger cohort study of the vascular phenotype of SSc. The criteria for inclusion in the study in the SSc cohort were diagnosis of SSc according to the criteria of the American College of Rheumatology , and age above 18 years old. SSc was classified as limited cutaneous (lcSSc) or diffuse cutaneous SSc (dcSSc) according to the criteria of LeRoy et al. . Exclusion criteria were cigarette smoking, diabetes mellitus, hypercholesterolemia, or any associated severe disease (cancer, cardiac and pulmonary failure, myocardial infarction, angina pectoris). Furthermore, patients receiving statins, nitrates or non-steroidal anti-inflammatory drugs were excluded. All patients were asked to discontinue any vasodilator therapy given for Raynaud's phenomenon one week before inclusion and until the end of the study. Patients unable to discontinue vasodilator therapies during the study period were not included.
The onset of the disease was defined as the first occurrence of symptoms of SSc except for RP. Digital pitting scars, esophageal dysfunction and RP were diagnosed clinically. Skin thickness was quantified using the modified Rodnan skin score . The diagnosis of pulmonary fibrosis was suspected on the basis of clinical data and systematic radiographs, and confirmed in all cases by computed tomography scans.
Primary RP was diagnosed according to the criteria of LeRoy and Medsger , including a normal nailfold capillaroscopy, the lack of antinuclear antibodies, no digital pitting scar and the lack of clinical symptoms of connective tissue disease.
This was a descriptive monocentric controlled study. All subjects gave informed written consent. The study was approved by the institutional review board of Grenoble University Hospital, France, on January 2004. Eligibility criteria and clinical status were assessed, and instructions for vasodilator therapy withdrawal were given to each subject. Twenty patients suffering from SSc were recruited from the Vascular Medicine Department. When a patient with SSc was enrolled, they were matched (sex and age ± 5 years) with a patient with Raynaud's phenomenon and a healthy volunteer. All patients with Raynaud's phenomenon and healthy volunteers were recruited through local newspaper advertisements.
All subjects arrived at the Clinical Research Center in Grenoble University Hospital between 8 a.m. and 9 a.m. in a fasting state, where the following measurements were performed within one day in a quiet room with a stable ambient temperature. After clinical examination, subjects were placed in a supine position with both forearms resting at heart level. Blood pressure and heart rate were determined and electrocardiography was performed, followed by laser Doppler measurements on the left arm. Venous blood samples were taken at fast for blood lipids and plasma glucose determination either before or after the laser Doppler measurements. Thereafter, subjects underwent echocardiography. Patients with SSc underwent pulmonary function testing.
Given the altered response to local heating we observed in SSc, we also tested in a pilot study whether this altered response was specific to this connective tissue disease. We studied 10 consecutive subjects with rheumatoid arthritis (RA) in the Rheumatology Department (Grenoble University Hospital, Grenoble, France) and 10 sex and age-matched (± 5 years) patients with primary Raynaud's phenomenon in the Inserm Clinical Research Center (Grenoble University Hospital, Grenoble, France), from April 2005 to May 2005. All subjects gave informed written consent. The study was approved by the institutional review board of Grenoble University Hospital, France, on April 2005. The criteria for inclusion in the study in the RA group were diagnosis of RA according to the American College of Rheumatology , and age above 18 years old. The criteria for inclusion in the study in the primary RP group were the same as in the main study. All subjects underwent thermal hyperhemia testing using the methodology detailed below.
Laser Doppler measurements
Cutaneous blood flow was measured using a laser Doppler flowmeter (PeriFlux System 5000, Perimed, Järfälla, Sweden). A laser probe (PR457) was attached to the distal pad of the third left finger and left in place during all the laser Doppler measurements. Data from the laser Doppler flowmeter were interfaced to a personal computer through a converter using Perisoft® data acquisition software.
Laser Doppler blood flow was recorded in mV, which are directly related to blood flow in the microcirculation of the surface tissue. Red blood cell flux values were divided by mean arterial pressure to yield a value of cutaneous vascular conductance expressed as mV/mm Hg. The expression of data in this manner takes into account any changes in blood flow due to change in blood pressure and also better reflects absolute changes in skin blood flow.
Following 30 minutes of rest, the hyperhemia was studied in the following sequence: post-occlusive hyperhemia with a 20 minute recovery period; sublingual nitroglycerin challenge with a 30 minute recovery period; and thermal hyperhemia. The recovery periods, determined in pilot experiments, were such that the cutaneous conductance returned to baseline values within these periods.
After 10 minutes of rest, to allow for the measurement of baseline cutaneous conductance, digital blood flow was occluded for 5 minutes by inflating a cuff placed on the left arm to 50 mm Hg above the systolic blood pressure. The cuff was then released and the flow responses were recorded. The amplitude of the response was determined by the peak hyperhemic conductance, expressed as an absolute value (mV/mm Hg). The kinetics of the response were determined by the time to peak hyperhemia and duration of hyperhemia, expressed in seconds.
Endothelium-independent vasodilation was tested 20 minutes later, following blood pressure and heart rate measurements. A single high dose of sublingual nitroglycerin (0.4 mg) was given. Digital skin blood flow was continuously recorded. The maximal effect was measured as the mean signal over a 1 minute period 4 minutes after nitroglycerin administration, similar to what is used for vasodilation of the brachial artery . The amplitude of the response was determined by the 4 minute peak conductance, expressed as mV/mm Hg.
We measured microvascular response to local heating 30 minutes later. The PR457 laser probe was heated to 42°C for 30 minutes and then to 44°C for 5 minutes. Laser Doppler flow measured over the first 30 minutes is characterized in healthy controls by an initial peak within the first 10 minutes followed by a nadir and a final rise to a second peak that continues as a sustained plateau. Maximal skin blood flow is achieved by heating to 44°C. The amplitude of the response was determined by the 10 minute thermal peak, 10–30 minute thermal peak, and 44°C thermal peak conductances, expressed as absolute values (mV/mm Hg). The maximal effects were measured as the mean signal over a 1 minute period for the 10 minute thermal peak, and as the mean signals over a 3 minute period for the 30 minute and 44°C thermal peaks. In subjects without a clear-cut initial peak, the maximal value of the first 10 minutes was measured as the mean signal over a 1 minute period, corresponding to the highest mean within the 10 minute window. The kinetics of the response were determined by the time to first thermal peak, and the time to second peak when available. The time to first peak was determined from the onset of the probe heating to the first peak, or to the maximal value when no clear plateau was observed.
Reproducibility of laser Doppler measurements
Reproducibility was tested on 20 healthy subjects for the thermal hyperhemic response, and on 10 healthy subjects for the post-occlusive hyperhemic response. Post-occlusive hyperhemia and thermal hyperhemia were performed as detailed above. Each examination was repeated 1 day after the end of the first series on the same subject. For thermal hyperhemia, the median absolute difference for the time to first thermal peak was 13 s (10th-90th percentile: 4–60) for a median of the means of 152 s (105–233). The median absolute difference for the 10 minute thermal peak was 4.5 mV/mm Hg (0.3–46) for a median of the means of 59 (20–81). For the post-occlusive response, the median absolute difference for the time to peak hyperhemia was 20 s (5–40) for a median of the means of 44 s (23–74). The median absolute difference for the peak hyperhemic conductance was 2 mV/mm Hg (0.5–9) for a median of the means of 46 (26–59). The coefficient of correlation for the time to first thermal peak and the 10 minute thermal peak was 0.89 and 0.65, respectively. The coefficient of correlation for the time to peak hyperhemia and peak hyperhemic conductance was 0.56 and 0.94, respectively. As correlation coefficients are poor indicators of reproducibility, Bland and Altman plots were constructed to measure the agreement between both measures. For the four measures, more than 95% of the differences were less than two standard deviations, and neither proportional error nor systematic errors were detected.
The mean time to first thermal peak was 154 s (standard deviation 56) in the 20 healthy controls involved in the repeatability study. Sample size calculations were based on the main objective, that is, to detect a difference in the time to first thermal peak of at least 60 s between groups, with α = 0.05 and power (1-β) = 0.9.
Quantitative data were analyzed with the following nonparametric statistical methods: Kruskal-Wallis analysis of variance; Mann-Whitney test for between groups comparisons; Wilcoxon test for paired analysis; and Spearman rank correlation test for the relationship between quantitative variables. Proportions were compared by using Chi2 tests or Ficher's exact test when appropriate. P-values less than 0.05, corrected by Bonferroni's method for multiple comparison, were considered significant. All quantitative data are expressed as the median, 10th and 90th percentiles. Qualitative data are expressed as number and percentage.