Article Text
PDF
Abstract
Objective The authors prospectively determined: (1) the specificity and sensitivity of dual energy CT (DECT) for gout; and (2) the interobserver and intraobserver reproducibility for DECT urate volume measurements.
Methods Forty crystal-proven gout patients (17 tophaceous) and 40 controls with other arthritic conditions prospectively underwent DECT scans of all peripheral joints using a gout protocol that color-codes the composition of tissues. A blinded radiologist identified urate deposition to calculate specificity and sensitivity of DECT for gout. Inter-rater volumetric reproducibility was determined by two independent radiologists on 40 index tophi from the 17 tophaceous gout patients using automated software.
Results The mean age of the 40 gout patients was 62 years, the mean gout duration was 13 years and 87% had a history of urate-lowering therapy (ULT). The specificity and sensitivity of DECT for gout were 0.93 (95% CI, 0.80 to 0.98) and 0.78 (0.62 to 0.89), respectively. When the authors excluded three gout cases with unreadable or incomplete scans, the sensitivity was 0.84 (95% CI, 0.68 to 0.94). The urate volumes of 40 index tophi ranged from 0.06 cm3 to 18.74 cm3 with a mean of 2.45 cm3. Interobserver and intraobserver intraclass correlation coefficients for DECT volume measurements were 1.00 (95% CI, 1.00 to 1.00) and 1.00 (95% CI, 1.00 to 1.00) with corresponding bias estimates (SD) of 0.01 (0.00) cm3 and 0.01 (0.03) cm3.
Conclusions These prospective data indicate high reproducibility of DECT urate volume measures. The specificity was high, but sensitivity was more moderate, potentially due to frequent ULT use in our patients.
Statistics from Altmetric.com
Introduction
The overall disease burden of gout is substantial and growing.1,–,4 However, there is no established imaging tool for accurate diagnosis and reliable follow-up of gout patients. Plain radiographs are not sensitive in diagnosing gout and often appear normal until the late stages of the disease.5 Although ultrasound images had previously been considered user-dependent, a recent retrospective study (n=23 gout patients) has reported that a ‘double-contour’ sign and presence of tophi could be very specific for urate deposition.6 Another study has reported that the inter-rater reproducibility of tophus diameter or volume by ultrasound may not be reliably high (ie, intraclass correlation range, 0.71–0.83).7 MRI for tophi are not specific for urate crystals8 and may lack reproducibility between readers as a potential tool to quantify tophus volume.9
A specific display algorithm based on the chemical composition of uric acid by dual energy CT (DECT) scans may provide a useful imaging tool for gout.10 Unlike conventional CT scans, DECT has been shown to specifically identify uric acid renal stones and colour-differentiate them from other types of renal stones both in vivo and in vitro.11 In a proof-of-concept retrospective study involving 20 patients with tophaceous gout and 10 controls, we have demonstrated that DECT scans produce obvious discriminating colour displays of urate deposits (100% sensitivity and 100% specificity) and the apparent colour-coded information allows for tophus urate volume quantification using a computerised process.10 Another initial retrospective evaluation (based on 12 crystal-proven gout cases and 19 crystal-negative controls) reported DECT's sensitivity of 100% and specificity of 89% (reader 1) and 79% (reader 2) for the diagnosis of gout.12 Neither study has evaluated the reproducibility of DECT tophus urate volume measurements, which is important for its potential utility as a quantitative follow-up tool.10 ,12
These data called for prospective investigation of the utility of DECT as a potentially accurate and reliable imaging tool in the diagnosis and follow-up of gout patients. To address these issues, we prospectively determined (1) the specificity and sensitivity of DECT for prevalent gout and (2) its interobserver and intraobserver reproducibility for urate volume measurements.
Methods
Study patients
We prospectively recruited 40 consecutive patients with crystal-proven gout (17 tophaceous) who agreed to undergo DECT scans between 1 December 2009 and 31 June 2011. Similarly, 40 consecutive control patients with other arthritic conditions prospectively underwent DECT (19 with rheumatoid arthritis, 13 with psoriatic arthritis, six with osteoarthritis, one with ankylosing spondylitis and one with undifferentiated inflammatory arthritis). After obtaining informed consent, a rheumatologist evaluated all study patients for history, physical examination and relevant data collection immediately before their DECT scans as described below. The current study was approved by the University of British Columbia Behavioural Research Ethics Board.
DECT scanning and evaluation of validity
DECT scans were performed at the Radiology Department of Vancouver General Hospital using a dual energy gout colour-coding protocol that specifically assessed the chemical composition of the material (ie, uric acid coloured in green, calcium coloured in blue).11 Patients received a DECT scan (SOMATOM Definition Flash, Siemens Healthcare, Forkheim, Germany) of all peripheral joints (ie, elbows, wrists, hands, knees, ankles and feet). The Siemens Definition scanner is equipped with two x-ray tubes and two corresponding detectors, which are angled at 95° to one another on a gantry. The two x-ray tubes are capable of rotating simultaneously at different energies (eg, 80 and 140 kV). This feature allows for the simultaneous acquisition of two data sets, essentially eliminating errors due to misregistration of data or patient movement between acquisitions. The two data sets are loaded on a multimodality workstation where a material decomposition algorithm allows accurate characterisation of uric acid. Attenuation values below the black line in figure 1 are rendered green (uric acid) and those above the black line are rendered blue (calcium) (figures 1, 2). The following scanning parameters were used: tube A, 140 kV/55 mA; tube B, 80 kV/243 mA; and collimation 0.6 mm reconstructed to 0.75 mm thick slices.
Three material decomposition display algorithm of DECT scans for gout. This algorithm separates the chemical composition of compounds based on their differential attenuation of the 80 kV and 140 kV x-ray beams. By applying the two different energy levels, three materials (ie, uric acid, calcium and soft tissue) can be accurately differentiated from one another based on their differential absorption levels of the x-ray beams. Material above the black line is colour-coded in blue, indicating calcium (ie, compound with high atomic weight number), and material below the black line is colour-coded in green, indicating uric acid (ie, compound with a low atomic weight number).
Dual energy CT images of uric acid deposits in various anatomical locations: (A) hand and wrist, (B) elbow, (C) knee, (D) feet. With the application of the three material decomposition algorithm, uric acid deposits are depicted in green, whereas calcium in bone is depicted in blue (trabecular bone in pink).
A board-certified radiologist blinded to the diagnosis determined the presence of urate deposition to calculate the specificity and sensitivity of DECT scans for gout.
Evaluation of reproducibility and DECT tophus urate volume measurement
We identified 40 index tophi from the 17 tophaceous gout patients for urate volume measurements. Two independent musculoskeletal radiologists quantified the urate volume of each index tophus using a computer automated procedure for inter-rater reproducibility. The same scans were scored twice for intra-rater reproducibility.
Using a dedicated automated volume assessment software program, volumes of uric acid deposition were measured. This process detects and quantifies only uric acid crystal elements, as opposed to whole organised tophi that include both uric acid crystal and non-crystal components such as cellular and fibrous components, which are included in other proposed methods such as MRI,9 tape measure13 and Vernier calliper techniques.14 The automated volume software allows radiologists to circle the entire bodily region (eg, forefoot) to determine total urate volume rather than having to manually locate and trace the tophus border, enabling a high degree of reproducibility (figure 3E).
Volume-rendered images of DECT scans in various locations. (A) Hand/wrist view displaying numerous large urate deposits (depicted in green) along the metacarpophalangeal and proximal interphalangeal joints (arrows). (B) Elbow joint view showing large urate deposits in the olecranon bursa of the elbow (arrow). (C) Knee joint view displaying urate deposits in the patellar tendon and prepatellar bursa (arrows). (D) Foot/ankle views illustrating large urate deposits at the metatarsophalangeal joints, midfoot and along the left Achilles tendon (arrows). (E) Demonstration of automated tophus urate volume calculations of the metatarsophalangeal joints based on circling each entire forefoot area.
Radiation safety consideration
The radiation dose of DECT scans for peripheral joints was optimised using phantom models in collaboration with our institution's medical physicist and was estimated to be no greater than that of the corresponding single energy CT scan. DECT's radiation dose was calculated to be 0.5 mSv per region scanned (eg, 0.5 mSv for both hands and wrists, which were scanned together). The total dose for all scanned peripheral joints in a given patient ranged from 2 to 3 mSv, which corresponds to the average annual natural background radiation dose (2.4 mSv).15
Statistical analysis
We calculated the specificity and sensitivity of uric acid deposition by DECT (coded green) with 95% CI. DECT positive was defined by the presence of green urate deposition in any of the scanned peripheral joints (ie, hands/wrists, elbows, feet/ankles and knees). Intraobserver and interobserver reproducibility was assessed by intraclass correlation coefficient and limits of agreement analysis (Bland and Altman). We compared the mean differences with zero difference by means of a one-sample paired t test. All tests used a two-tailed significance level of 0.05. Analysis was performed using the STATA software package (STATA Corporation, College Station, Texas, USA).
Results
Patient demographics and clinical characteristics at the time of the DECT study are summarised in table 1. The mean age of the 40 gout patients was 62 years and 35 were male subjects (88%). The mean duration of gout was 13 years and the mean serum uric acid level at the time of DECT scan was 6.3 mg/dl. Thirty-eight patients (95%) had experienced an acute gout attack in the year prior to the study and 46% had an attack within the 4 weeks prior to their DECT scan visit. Notably, 35 (88%) had been treated with urate-lowering therapy. Reflective of the demographic characteristics of their underlying rheumatic conditions, mean age of controls was 53 years and 13 (33%) of them were male subjects (table 1).
Demographic and clinical characteristics of study patients
While all 40 control patients and 37 gout patients had complete target area DECT images available, three gout cases had missing target area images due to motion artefacts or image storage with a wrong kernel use. Representative DECT scan images from gout patients are illustrated in figures 2, 3.
Specificity and sensitivity of DECT urate deposition for the diagnosis of gout
Based on all study participants (n=80), the specificity and sensitivity of DECT for the presence of gout were 0.93 (95% CI, 0.80 to 0.98) and 0.78 (0.62 to 0.89), respectively. When we excluded the three gout cases for whom all regions were not available for evaluation, the sensitivity was 0.84 (95% CI, 0.68 to 0.94).
Five of the six false negative gout patients were on allopurinol and had serum uric acid levels <6 mg/dl. Of the three false positive cases, one had rheumatoid arthritis with serum uric acid of 7.5 mg/dl and DECT urate deposits in the knees and feet. However, the other two (one with rheumatoid arthritis and the other with osteoarthritis) had normal serum urate levels and small urate deposits (<2 mm in diameter), both in the feet.
Reproducibility of urate volume measurements within tophi
The urate volumes of 40 index tophi ranged from 0.06 cm3 to 18.74 cm3 with a mean of 2.45 cm3. Of these 40 tophi, 22 (55%) were located in the feet and ankles, 12 (30%) in the knees and elbows and six (15%) in the hands and wrists. There was no statistically significant difference between readers in mean tophus urate volumes for all locations or for individual locations (all p values >0.36) (table 2). The mean difference (ie, bias) (SD) in tophus urate volumes between readers was 0.00 cm3 (0.02) for all locations, which was similar when we analysed individual locations (table 2). A plot of the differences in tophus urate volume between readers as a function of mean volume of the respective urate deposit (figure 4) showed that measurement differences between readers were similar regardless of tophus urate volume. The intraclass correlation coefficient between readers was 1.00 (95% CI, 1.00 to 1.00) for all or individual locations and the mean (SD) of average percent difference between readers was 1% (0%). There was no statistically significant intrareader difference in mean tophus urate volumes for all locations or for individual locations (p value=0.13). The mean difference (ie, bias) (SD) in tophus urate volumes within readers was 0.01 (0.03) for all locations and the corresponding intraclass correlation coefficients were 1.00 (95% CI, 1.00 to 1.00).
Bland–Altman plots of difference between readers. Differences in tophus urate volumes measured by two independent readers plotted as a function of mean tophus urate volume. The red solid horizontal line represents the mean difference (ie, bias) in volumes measured by the two readers. The dashed horizontal lines represent the observed 95% limits of agreement.
Summary of inter-rater reproducibility analysis
Discussion
In this prospective validation study involving established gout patients with urate crystal confirmation and other arthritic conditions as controls, we found that the specificity of DECT scans for the presence of gout was high and sensitivity was moderate. Our prospectively derived specificity estimate (93%) appeared higher than the recently reported retrospective estimates (79% and 89%) based on 19 controls,12 whereas our prospective sensitivity estimates (78% and 84%) were lower than the two retrospective findings (100%) based on 2010 and 12 cases of gout.12 It is conceivable that our sensitivity estimates may have been influenced by the frequent history of urate lowering therapy use in our gout patients (88%). We also found that DECT tophus urate volume measures are highly reproducible, supporting DECT as a reliable imaging tool for quantifying tophus urate volumes in gout patients.
Previous studies have evaluated the reproducibility of various tophus quantification methods, including tape measure (area),13 Vernier calliper (length),14 ultrasound (diameter and volume),7 MRI (volume)9 and conventional CT scans (volume).14 ,16 Reported interobserver intraclass coefficients (bias (ie, average difference between readers)) associated with tape measures (0.92 (150 mm2)) and ultrasound (0.83 (1.06 mm))7 ,13 tended to be inferior to those associated with the other methods (0.99 (0.45 mm) for Vernier calliper, 0.98 (0.89 cm3) for MRI and 0.99 (0.0652 cm3) for conventional CT).9 ,14 In comparison, DECT's interobserver intraclass coefficient was 1.00 with a bias of 0.00 cm3, suggesting that DECT volume measures are extremely reproducible, perhaps more so than other tophus measurement methods. This high level of reproducibility likely stems from DECT's reliance on automated computerised procedures that detect and quantify only uric acid crystal elements (thus, not including non-crystal components of tophi) in an region (figure 3E) rather than other methods (eg, MRI)9 ,14 that require manual localisation and border tracing of urate deposition to determine tophus volume.
DECT scans carry several potential advantages as a follow-up imaging tool over MRI, which was recently used to quantify tophus size in a randomised trial.17 Nevertheless, unlike MRI, DECT involves radiation and its exposure level is the same as conventional CT. The DECT radiation dose has been estimated to be 0.5 mSv of dose per scanned region at our institution and the total amount for all peripheral joints per patient ranges from 2 to 3 mSv, which corresponds to the annual global per caput average dose due to natural radiation sources (2.4 mSv).15 For comparison, a relevant organ radiation dose for screening mammography, a widely-accepted public health cancer prevention measure, is estimated to be 3 mSv.18 Furthermore, the narrow collimation beam achieved with CT provides minimal scatter, unlike that created by the plain radiographs, further limiting radiation exposure to other areas.
Potential limitations of our study and future research directions deserve comment. The demographics of our gout patients and controls were reflective of those of underlying arthritic conditions and, thus, gout patients were older and more often male subjects. While our additional analysis of our study patients indicated no independent association between these demographic factors and DECT positivity, future studies could examine whether DECT's diagnostic accuracy varies in different demographics. As our validation study involved crystal confirmed patients regardless of treatment duration or urate lowering therapy, it would be valuable to prospectively evaluate whether the sensitivity differs among untreated gout cases. Furthermore, studying new-onset gout cases would also be relevant to medical decision making in a common clinical context of acute inflammatory arthritis. Future validation studies should also involve more than one centre reviewing the DECT images obtained using different machines. Finally, evaluating DECT as a follow-up tool for group discrimination and change sensitivity in the actual care of gouty patients, specifically in following the treatment response to urate-lowering agents, would also be informative.
In conclusion, these prospective data indicate that DECT tophus urate volume measures are highly reproducible, supporting DECT as a reliable imaging tool for following urate volumes in gout patients. The specificity of DECT scans for the presence of gout was high, but sensitivity was more moderate, which could be due to frequent urate-lowering therapy use in our gout patients. Future research should examine whether the sensitivity differs among untreated, new-onset gout cases.
Acknowledgments
The authors thank the rheumatology and radiology faculty members of the University of British Columbia for providing patient information and valuable comments for the study.
References
Footnotes
Funding This study was supported by investigator initiated research grants from Takeda Pharmaceuticals. The sponsor had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript.
Competing interests Dr Choi has served on advisory boards for Takeda Pharmaceuticals, URL Pharma and Savient Pharmaceuticals.
Patient Consent Obtained.
Ethics approval Approval provided by the University of British Columbia Behavioural Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.