ARFI Imaging for Differentiation of Thyroid Nodules

Abstract

Background

Acoustic Radiation Force Impulse (ARFI)-Imaging is an ultrasound-based elastography method enabling quantitative measurement of tissue stiffness. The aim of the present study was to evaluate sensitivity and specificity of ARFI-imaging for differentiation of thyroid nodules and to compare it to the well evaluated qualitative real-time elastography (RTE).

Methods

ARFI-imaging involves the mechanical excitation of tissue using acoustic pulses to generate localized displacements resulting in shear-wave propagation which is tracked using correlation-based methods and recorded in m/s. Inclusion criteria were: nodules ≥5 mm, and cytological/histological assessment. All patients received conventional ultrasound, real-time elastography (RTE) and ARFI-imaging.

Results

One-hundred-fifty-eight nodules in 138 patients were available for analysis. One-hundred-thirty-seven nodules were benign on cytology/histology, and twenty-one nodules were malignant. The median velocity of ARFI-imaging in the healthy thyroid tissue, as well as in benign and malignant thyroid nodules was 1.76 m/s, 1.90 m/s, and 2.69 m/s, respectively. While no significant difference in median velocity was found between healthy thyroid tissue and benign thyroid nodules, a significant difference was found between malignant thyroid nodules on the one hand and healthy thyroid tissue (p = 0.0019) or benign thyroid nodules (p = 0.0039) on the other hand. No significant difference of diagnostic accuracy for the diagnosis of malignant thyroid nodules was found between RTE and ARFI-imaging (0.74 vs. 0.69, p = 0.54). The combination of RTE with ARFI did not improve diagnostic accuracy.

Conclusions

ARFI can be used as an additional tool in the diagnostic work up of thyroid nodules with high negative predictive value and comparable results to RTE.

 

Acoustic Radiation Force Impulse (ARFI)-Imaging

 

Velocities of ARFI measured in thyroid nodules and tissue are shown in Table 3. In 5 patients no measurement in the healthy thyroid gland was possible due to multinodular goiter. While no significant difference in median velocity was found between healthy thyroid tissue and benign thyroid nodules (p =0.068), a significant difference was found between healthy thyroid tissue and malignant thyroid nodules (p =0.0019), as well as between benign and malignant thyroid nodules (p = 0.0039), respectively.

The median success-rate of ARFI-measurement (number of valid measurements divided by the number of all measurements performed) was 100% in the healthy thyroid (mean: 99+/-2%, range: 91–100%), and 100% in thyroid nodules (mean: 92+/-17%,range: 9–100%). The lower mean success-rate for nodules accounts for the upper measurement limit of 8.4 m/s above which values were displayed as ‘‘x.xxm/s’’ and therefore counted as unsuccessful measurement.

AUROC for ARFI of the thyroid nodule for the diagnosis of malignant thyroid nodules was 0.69 [95-CI: 0.53;0.85] (p = 0.0043). The optimal cut-off with the highest sum of sensitivity and specificity (Youden cut-off) for ARFI-measurement in thyroid nodules was 2.57 m/s (Table 2). AUROC for the ratio of ARFI in  the nodule and healthy thyroid tissue for the diagnosis of malignant thyroid nodules was 0.71 [95-CI: 0.56;0.85] (p = 0.0025). The optimal cut-off (Youden cut-off) for ARFI-ratio was 1.57 m/s (Table 2). No significant difference was found between AUROC of ARFI of the nodule and ARFI-ratio (p.0.20). Details are shown in Table 2 and Figure 3.

 

Intra-observer variability expressed as the mean standard deviation of 10 measurements at one location was 0.46 within all thyroid nodules, and 0.21 within healthy thyroid tissue. It was higher in malignant nodules with 0.94 as compared to benign nodules with 0.39.

Discussion

RTE has become a well evaluated clinical tool enabling the determination of tissue elasticity using ultrasound devices. RTE is a qualitative elastography method evaluating changes in ultrasound pattern during strain and stress of direct or indirect tissue compression. A recent meta-analysis reported a sensitivity and specificity for RTE for the diagnosis of malignant thyroid nodules of 92%, and 90%, respectively [12]. Methods to quantify the colour coded images revealed by RTE were developed using strain value and ratio and histograms with the aim of reducing intra- and interobserver variability [27]–[29]. Nevertheless, besides a lot of promising study results two recent studies have challenged the usefulness of RTE in clinical practice by reporting no additional value as compared to qualified B-mode ultrasound [13], [30].

Quantitative elastography was well evaluated for the diagnosis of liver fibrosis with most studies evaluating transient elastography (FibroScan, Echosens, Paris) [31]. Hereby, a mechanical wave is send into the liver and the velocity of shear waves within the liver is measured. However, it was only developed for measurement in liver tissue. Recently, other quantitative elastography methods were developed, which are integrated in conventional ultrasound systems and can be performed in all solid organs [32], [33]. These quantitative methods also send a mechanical or acoustic wave into the tissue and measure the velocity of shear waves; the stiffer the tissue is, the faster the shear waves propagate. Only one previous study with 146 nodules from 93 patients evaluated shear wave elastography with SuperSonic Imaging (Aixplorer, Aixen Provence, France) and reported a sensitivity of 85%, and a specificity of 94% for the diagnosis of malignant thyroid nodules [34]. In a recently published pilot study [14] the feasibility of Acoustic Radiation Force Impulse (ARFI)-imaging to measure thyroid tissue and thyroid nodules was shown. To our knowledge, no previous study has compared the well evaluated qualitative RTE and the novel quantitative elastography methods for the differentiation of thyroid nodules. The results of the present study show comparable results for RTE and ARFI-imaging for the differentiation of benign and malignant thyroid nodules. Sensitivity and specificity of RTE in the present study was lower than in the published meta-analysis on RTE [12] with sensitivity of 76% vs. 92%, and specificity of 72% vs. 90%. However, the results or RTE in the present study were higher than in the study of Moon et al. [30] evaluating 703 nodules in 676 patients with sensitivity of 75% vs. 65%, and specificity of 72% vs. 58%. The results and these discrepancies again might be explained by the qualitative and operator-depending procedure of RTE. The advantage of ARFI-imaging is that the same acoustic wave is send into the tissue independent of the examiner pressing the button to start measurement, while for RTE the examiner needs to perform small compressions to the tissue which may vary. In addition, RTE determines tissues elasticity in relation to surrounding tissue, whereas ARFI is a quantitative method measuring the velocity of shear waves within a ROI.

The combination of RTE with ARFI-imaging improved specificity for the diagnosis of malignant thyroid nodules from 72% (RTE alone) to 92% (combination of both), but reduced sensitivity from 76% to 48%, respectively. Both methods revealed an excellent negative predictive value for excluding malignant thyroid nodules with 95% for RTE alone, and 93% for ARFI alone. The combination of both methods did not further improve NPV.

A possible clinical algorithm could be to use primarily one elastography method in combination with FNAB to exclude malignancy of a thyroid nodule and perform follow-up examinations in patients with benign FNAB and benign criteria on RTE or ARFI. However, both methods might be useful in combination if FNAB reveals benign cytology, but one elastography method shows criteria of malignancy. If then both methods (RTE and ARFI) report values in the range of malignancy, than operation could be advised despite the benign cytology. Nevertheless, of course B-mode ultrasound criteria must be included in such an algorithm. Further larger studies are necessary to find an optimal algorithm of B-mode ultrasound, qualitative and quantitative elastography and FNAB to optimize the work up of thyroid nodules.

The present study has the following limitations:

The reference standard was cytology only in 94/158 (59.5%) nodules with benign cytology. However ultrasound examination after 6 months did not show growth of nodule size as a sign of benign lesions. Nevertheless, false-negative cytology may have existed. Histology was the only excepted reference method for the diagnosis of malignant thyroid nodules. The malignant nodules were predominantly papillary carcinoma which might limit the diagnostic utility to this entity.

Cystic lesions without at least 5×5 mm of solid parts of the nodules were excluded from the present study, since ARFI –ROI measures 5×5 mm and cystic lesions produce artefacts on RTE mimicking hard tissue. Therefore, the results of the present study cannot draw any conclusion concerning the value of ARFI for predominantly cystic lesions.

The guidelines for clinical practice for the diagnosis and management of thyroid nodules of the American Association of Clinical Endocrinologists (AACE), Associazione Medici Endocrinologi (AME) and the European Thyroid Association recommend that suspicious thyroid nodules smaller than 10 mm should be assessed by FNAB [5]. However, in the present study, only 22 nodules with 5–10 mm in size were included, which was too small to perform a subanalysis. A recent study demonstrated, that RTE can be performed in thyroid nodules of 3–10 mm in size and is suitable for the diagnosis of microcarcinoma of the thyroid gland [35]. Future studies should evaluate the value of ARFI and the combination of ARFI with RTE in thyroid nodules smaller than 10 mm.

The intra-observer variability expressed as the mean standard deviation of 10 measurements at one location was 0.46 within thyroid nodules, and 0.21 within healthy thyroid tissue. Especially in malignant thyroid nodules it was as high as 0.94. A reason might be that many measurements in malignant nodules resulted in “x.xx m/s” if the value exceeded the upper detection limit of 8.4 m/s. In these cases more than 10 measurement attempts were made to reach 10 numeric values. A software optimization increasing the velocity detection at velocities exceeding 8.4 m/s are needed to overcome this limitation.

In summary, the present study demonstrates comparable results for the novel quantitative Acoustic Radiation Force Impulse-Imaging as for the well evaluated qualitative real-time elastography for the differentiation of thyroid nodules. The combination of both methods did not significantly improve the diagnostic accuracy for the diagnosis of malignant thyroid nodules. Large multicenter studies are necessary to develop application algorithms for qualitative and quantitative elastography in clinical practice.