FLT-PET for the assessment of systemic sarcoidosis including cardiac and CNS involvement: a prospective study with comparison to FDG-PET

The last few years have seen an increased recognition of the importance in the detection of cardiac sarcoidosis (CS), accounting for up to 85% [1] of sarcoid deaths. In addition, neurosarcoidosis (NS) is associated with a 10% mortality rate [2] and present in 5–15% of cases [3]. For both NS and CS, post-mortem studies have revealed that only a fraction of cases are diagnosed antemortem with up to 39% and 50% of cases of CS and NS detected on autopsy [4].

2-deoxy-2-[¹⁸F]fluoro-d-glucose (FDG)-PET is useful in diagnosing and assessing the extent of active sarcoidosis, due to its ability to image inflammation. FDG-PET has been shown to be useful in identifying appropriate biopsy sites in patients with sarcoidosis as affected lymph nodes are often morphologically unremarkable and undetectable by other imaging modalities [5]. Furthermore, the degree of FDG uptake, particularly within lung parenchyma, has been shown to serve as an effective therapeutic target [6] with the degree of lung uptake decreasing following the initiation of therapy [7]; however, despite a near perfect sensitivity of 96–100% for pulmonary disease [8], the diagnostic accuracy for cardiac involvement is lower, with one meta-analysis reporting a sensitivity/specificity of 89%/78% [9]. Furthermore, extensive patient preparation is required to suppress physiological FDG uptake and non-diagnostic or false positive studies due to suboptimal suppression can be seen in 10–20% of cases [10]. The role of FDG-PET in assessing NS remains largely unstudied.

3′-deoxy-3′-[¹⁸F]fluorothymidine (FLT)-PET has recently been shown to be useful in the assessment of CS without requiring extensive pre-imaging preparation [11, 12]. This prospective study investigated the use of FLT-PET for the evaluation of sarcoidosis, including myocardial and neural involvement, with comparison to FDG-PET.

Imaging studies were performed in the following order: (1) cardiac rest perfusion scan; (2) whole-body FDG scan; (3) ECG-gated cardiac FDG scan. FLT studies were acquired in the following order: (1) whole-body FLT scan; (2) ECG-gated cardiac FLT scan. All images were acquired in 3D mode and reconstructed using an iterative technique. PET images were performed on a Discovery 690 or 600 PET/CT scanner (GE Healthcare, Waukesha, WI). FLT scans were performed within 14 days of perfusion and FDG scans.

Images were anonymized and reviewed in random order by two experienced nuclear medicine physicians (PM and DJ) blinded to any identifying information using HybridViewer PET-CT fusion software (Hermes Medical Solutions). CT findings were reviewed and used for anatomic localization and characterization of PET findings. Perfusion images and dedicated cardiac DG-PET and FLT-PET were reviewing using 4DM (INVIA Medical Imaging Solutions). Regions of interest (ROIs) were defined as spheres 1 cm in diameter. All reported SUVs consisted of SUV_Max measured within an ROI. Blood pool SUVs were measured in a similar fashion using an ROI within the ascending thoracic aorta. Individual lesions were defined as foci of radiotracer uptake significantly greater than background and visually distinct from the normal biodistribution of the radiotracer. Active sarcoidosis was defined as uptake of either FDG or FLT in lesions consistent with sarcoidosis. A third reader resolved differences in interpretation.

The gold standard for sarcoidosis was tissue diagnosis while the gold standard for cardiac involvement consisted of the Heart Rhythm Society criteria [14]. Clinical diagnosis was considered the gold standard for NS.

Descriptive statistics are presented as the number (percent) for categorical variables and mean (± standard deviation) for continuous variables. Normality was tested using the Shapiro–Wilk test. Differences between groups were compared using Student’s paired samples t test. Kappa (κ) statistics were calculated for inter-observer agreement of categorical findings on FLT- and FDG-PET scans while continuous variables were compared with the intra-class correlation coefficient (ICC) using a two-way random effects, absolute agreement, single rater model [15]. Interpretation of concordance measures was in keeping with the guidelines of Cicchetti et al. [16]. A p value of < 0.05 was considered statistically significant. All analyses were performed using SPSS v20 (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp).

In this prospective study, FLT-PET demonstrated excellent imaging and diagnostic properties for the evaluation of extra-cardiac sarcoidosis, as well as CS. FLT-PET was able to detect multiple NS lesions in a single subject with known neural involvement.

Limitations inherent in cardiac FDG-PET imaging are well-known and include extensive patient preparation, variable success in suppression of myocardial activity, and the existence of normal patterns of myocardial uptake complicating the interpretation of studies [10]. In particular, interpretation of the various cardiac uptake patterns is associated with only moderate inter-observer agreement [17]. Due to high, non-suppressible physiological uptake seen in the central nervous system (CNS), FDG-PET plays very little role in the evaluation of NS, with the extant literature limited to case reports. Because of these limitations, there is a need for a radiotracer which can accurately assess not only the lung parenchyma and lymph nodes, but also the myocardium and CNS, without the need for extensive patient preparation.

FLT offers multiple advantages, especially for the assessment of CS and NS. In particular, lack of physiological myocardial uptake obviates the need for extensive patient preparation resulting in increased patient convenience, and decreased risk of non-diagnostic studies due to patient non-compliance with pre-imaging preparation instructions. Unlike FDG, there are no known normal variants demonstrating non-pathological myocardial FLT uptake, simplifying cardiac interpretation while lowering the risk of false-positive studies. In particular, our subject with a normal variant (i.e., lateral wall FDG uptake) had an unremarkable cardiac FLT study. In our subject with NS, both FLT and FDG were able to detect paraspinal lesions; however, only FLT adequately demonstrated intracalvarial disease, due to a lack of background cerebral uptake. The high concordance in the distribution of lesions seen on FLT and FDG at key sites (i.e., heart, CNS, lung parenchyma), as well as for the overall number of organs involved, suggests a comparable performance to FDG disease staging. Furthermore, FLT effectively identified small pathological lymph nodes which could be used to identify active sites of disease for diagnostic biopsy. FLT showed a non-significant trend towards lower sensitivity for the detection of abdominal lymph nodes compared to FDG—this may have be related to the observation that, in many patients, involved lymph nodes are commonly seen around the porta hepatis and the high hepatic uptake seen on FLT may make identification of lymph nodes at the porta challenging; however, further studies will needed to confirm this. Although not shown in our study, FLT-PET could, in theory, be less sensitive than FDG-PET for the detection of sarcoidosis lesions within the liver and bone marrow due to the significant uptake normally seen in these organs; however, this may deemed acceptable in light of the improved assessment of the heart and CNS and the relatively limited clinical impact of hepatic and marrow involvement.

Our study was not designed to determine the utility of FLT lung uptake as a therapeutic target; however, one would expect that, in order to duplicate the success of FDG in this respect, there should at least be good concordance between FLT and FDG for the assessment of parenchyma involvement, which was present in our results. Additional studies will be needed to assess the effectiveness of FLT-PET in the assessment of therapeutic response.

Although FLT demonstrates a lower degree of uptake (SUV) in sarcoidosis lesions when compared to FDG, the difference was not clinically significant with lesions demonstrating similar signal to background with both tracers. After correction for background activity (i.e., blood pool) the difference in SUVs was not statistically significant. Linear regression analysis revealed a relatively low degree of correlation between FDG and FLT uptake compatible with our previously reported results [12]. This can be attributed to the fact that the degree of FDG uptake is non-specific and reflects all underlying metabolic processes while FLT uptake is specific to proliferative activity.

FLT activity is expected to represent granuloma burden while FDG reflects general inflammatory activity. Differences in uptake mechanisms between FLT and FDG could be useful for disease monitoring and therapy assessment. In particular, we speculate that the specificity of the uptake mechanism of FLT may be particularly well-suited for monitoring response to therapy; however, additional studies are needed.

It is unclear why a single subject (#10) showed discordant myocardial findings (positive on FDG, negative on FLT) while another subject (#8) showed discordant whole-body FLT and FDG findings (positive on FLT, negative on FDG). In both cases, these subjects were undergoing treatment at the time of imaging. It may be that the discordance was related to treatment effects. Also, it should be noted that the imaging findings in both cases were subtle and the discordances might be attributable to limited sensitivity (i.e., false-negative) and/or specificity (i.e., false-positive) for either FLT or FDG.

Our results can be compared to those of Norikane et al. [11] who conducted a retrospective study involving 20 subjects with newly diagnosed cardiac sarcoidosis and compared FLT and FDG-PET findings. They found that both FLT and FDG had high accuracy for cardiac and extra-cardiac thoracic sarcoidosis. There are, however, several important differences between our studies. Our study was prospective and used the HRS criteria as gold standard, compared to the JMHW criteria used by Norikane et al. Furthermore, subjects undergoing cardiac evaluation in the Norikane study had a high rate of inconclusive FDG studies (4/20) while all of our subjects’ studies were diagnostic. Also, Norikane limited their imaging to the thorax while we acquired near whole-body images, which enabled us to compare findings in extra-thoracic sarcoidosis (including intracalvarial NS). Importantly, the work by Norikane et al. did not incorporate cardiac perfusion imaging in their analysis—we have previously shown significant associations between the findings on FLT-PET and perfusion imaging [12]. Finally, none of their subjects had evidence of NS.

Some limitations of this study should be acknowledged. Our sample size was small, and lack of negative controls prevented us from calculating the accuracy for whole-body sarcoidosis. In order to remedy this, a larger prospective study is underway. In addition, all subjects were referred for suspicion of CS. This subgroup of subjects may not be representative of typical subjects with sarcoidosis and may have influenced the distribution of sarcoidosis lesions we observed in our study. Finally, there was only a single subject with a clinical diagnosis of NS, limiting our assessment of the utility of FLT-PET for neural involvement.

In summary, FLT-PET has the ability to accurately detect sarcoidosis lesions, including cardiac and CNS involvement, without the need for extensive patient preparation. Follow-up studies are ongoing in order to confirm these findings.