Includedpatient cohorts
Eighty patients diagnosed with Pso/PsA, 80 patients diagnosed with RA, and 28 healthy controls were selected from different cohort studies. We included patients with a definite diagnosis of Pso (diagnosed by a dermatologist), PsA (diagnosed by a dermatologist and a rheumatologist), or RA (diagnosed by a rheumatologist) and at least one FOI examination.
Exclusion criteria were observed psoriatic lesions or other obvious non-psoriatic on the hands, which were documented for each patient via illustration. Patients who were missing these illustrations were excluded. Furthermore, a glomerular filtration rate (GFR) below 60 ml/min, hyperthyroidism, breast feeding, pregnancy and age below 18 years were general exclusion criteria to perform the FOI examination (see Appendix for the flow chart concerning the patient recruitment).
Ethical committees in Germany gave the required ethical approvals for the respective studies (128/13 EK, 127/13 EK, EA1/025/1, EA1/193/10, EA1/269/13). All patients from the different studies were originally recruited in the department of rheumatology (i.e. outpatient clinic and day unit) of the Charité – Universitätsmedizin Berlin. All included patients got written and oral information about the study and signed the informed consent.
FOI method and examination
Ten seconds after start of the examination, the fluorescent dye Indocyanine green (ICG) is intravenously administered (0.1 mg/kg of the body weight). After stimulation of the dye by light of the near-infrared spectrum, a special charge-coupled-device (CCD) camera detects occurring light emission and presents it visually as enhancements in both hands. During the six minutes of examination, one image per second is recorded adding up to a cluster of 360 pictures in total. To assess the examination, there are different methods described. The most common method to analyze joint inflammation on FOI scans is the Berlin method, which classifies the image sequence into three phases. Depending on the signal in the fingertips, the signals in each phase are evaluated regarding intensity, size and shape and are classified into three grades (0-3) according to the semi-quantitative FOI Activity Score (FOIAS).(29, 36-38) Using this assessment, joints showing hypervascularization can be detected and thus, the activity of inflammation can be estimated.
Detecting subclinical skin enhancement
In order to detect potential subclinical skin enhancement, we developed a new semi-quantitative (0-3) score to describe the degree of the skin enhancement. The score concentrates on the first part of the image sequence (0-120 sec) since we noticed that the enhancements were only visible in this time frame.
The FOI scans were randomized and blinded, then evaluated according to our defined criteria using example images in an atlas as reference (Supplementary File). The images were scored by one reader (AS). In order to analyze the interreader reliability, a second reader (SO) scored a sample of 90 randomized and blinded cases.
These defined criteria had to be fulfilled before evaluation:
- at least 90% of the respective hand must be green flooded in the respective time frame
- the enhancement suspicious of subclinical skin enhancement was assumed to be localized on the back of the hand without any relationship to a joint, or blood vessel.
- the enhancement had to be at least yellow with red spots in intensity to be considered as such.
The color intensity was evaluated semi-quantitatively (grade 0-3) similar to the FOIAS (Figure 1):
- 0 = no signal enhancement
- 1 = enhancement varies from yellow to red and can reach red with yellow spots, red covers ≤50% of the enhanced/affected joint area,
- 2 = the signal intensity shows strong red colour intensity and can also include white signals, and white covers ≤50% of the enhanced/affected joint area,
- 3 = the signal intensity shows white colour intensity, and white covers >50% of the enhanced/affected joint area
To ensure that the detected skin enhancement did not show a flooding vessel, it was recommended to read the same sequence with the XiraView 3.7 "Temperature" filter. This is a particularly effective way of displaying vessels and it is possible to check the area in question for them. (Examples in Supplementary Figure S2)
In order to describe where the lesions were most frequently found, the back of the hand was divided into five regions (Figure 2):
- Region 1: below the 2nd metacarpal joint (MCP2) and next to MCP1, not touching any of them or the thumb base joint
- Region 2: below MCP3 down to the middle of the back of the hand
- Region 3: below MCP4 down to the middle of the back of the hand
- Region 4: below MCP5 down to the middle of the back of the hand
- Region 5: above the wrist joints up to the middle of the back of the hand
An FOI-based diagnosis was made based on the subclinical skin enhancement on the dorsum of the hand and enhancement of joints and tendons (Figure 3):
- Pso/PsA: subclinical skin enhancement and (in case of PsA) further enhancements above the joints (according to FOIAS)
- RA: No subclinical skin enhancement (but enhancement above the joints according to FOIAS)
- Healthy control: No subclinical skin enhancement (and no further enhancements above the joints according to FOIAS)
We further characterized the FOI patterns and sorted the scans into groups based on the assumed diagnosis (Pso/PsA, RA and healthy controls), which was compared with the physician’s diagnosis. The physician’s diagnosis was extracted as the final diagnosis in the last available physician’s letter.
To identify possible influences on our FOI findings, we tested potential influencing factors (e.g. CV risk factors, sex, disease duration) for both false positive and false negative results. False positive results refer to detected skin enhancement in patients who, according to our definition, should not show any skin enhancement (i.e. RA patients and healthy controls). Accordingly, false negative results are non-existing skin enhancements in Pso/PsA, since we hypothesized that all Pso/PsA patients should show subdermal skin enhancements.
Furthermore, we investigated potential association between CV factors and imaging findings. CV risk factors such as hypertension, metabolic syndrome or smoking status were recorded and correlated to the imaging results.
Statistical analysis
Sociodemographic and clinical characteristics were compared by analysis of variance for continuously distributed variables or a Chi-square test for categorical variables between the three groups Pso/PsA, RA and healthy controls. The concordance between the rate of FOI positive cases and the FOI diagnosis by one reader with the clinical diagnosis was determined. Based on the gold standard of the clinical diagnosis, the rates of true positive, true negative, false positive and false negative FOI results were calculated. Multinomial logistic regression analyses were conducted to model the association between the four groups true positive, true negative, false positive and false negative with clinical and sociodemographic characteristic as predictor variables. The reference group was set to positive classified patients in the multinomial logistic regression model. In addition, the association of clinical and sociodemographic variables with the outcome FOI enhancement was estimated by means of binary logistic regression model. The interreader agreement was investigated in 90 randomly selected and blinded patients between the ratings of the two readers AS and SO. Absolute agreement rates and the kappa coefficient were calculated to evaluate agreement. A p-value of < 0.05 was considered statistically significant. Statistical analyses were conducted with the statistical software STATA 12.1 (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP).