Reply to the Letter to the Editor: “Visualization of the saccule and utricle with non-contrast-enhanced FLAIR sequences”

by Hikaru Fukutomi, Thomas Tourdias (thomas.tourdias@chu-bordeaux.fr)

Visualization of the saccule and utricle with non-contrast-enhanced FLAIR sequences

Dear Editor-in-Chief,

We thank Drs Naganawa and Sone for their detailed comments on our article.

First, regarding T1 and T2 values of endolymphatic space (ELS) and perilymphatic space (PLS), it is important to mention that these structures are too small to allow direct quantification of both T1 and T2 values. However, a ROI in the center of the vestibule (likely within ELS) and a ROI in the periphery of the vestibule (likely within PLS) provided important data that are not just hypothesis. Indeed, the null point in standard 3D-FLAIR is mainly T1-driven. The fact that ELS and PLS have identical nulling point means that these structures have identical / close T1 values (Fig 3c). Furthermore, it is known that the T2Prep module decreases T1 contrast while increasing T2 contrast, which is why separation of the nulling point with a T2Prep module indicates difference in T2 values that are likely to be too subtle to be captured with standard FLAIR. Therefore, we would like to clarify the fact that Fig 2 should help the reader to understand how the signal is theoretically supposed to behave depending on the T1 and T2 values of fluids, but Fig 3 provides direct experimental data pointing to differences of relaxation times of PLS and ELS. We fully agree with Drs Naganawa and Sone that the underlying differences in fluid composition responsible for such differences in T1/T2 values is still to be elucidated.

Second, regarding differences in shape of the vestibule that Drs Naganawa and Sone noted in Fig 3b, it is important to keep in mind that this particular experiment has been conducted at intermediate resolution (1 x 1 x 1.5 mm) that was mandatory to keep a reasonable scan time while repeating 16 acquisitions (for 16 different TI values). Therefore, it is not fully relevant to make detailed anatomical analyses on these images that are likely affected by partial volume averaging. The reader could observe from the curves in Fig 3e that the signal from the PLS is lower at TI = 224 ms than at TI = 5000 ms which is likely to explain that the periphery of PLS is not visible at TI = 224 ms because of partial volume effect with the adjacent bone. Similarly, the cochlea size increased with longer TI due to higher signal.

Third, Drs Naganawa and Sone raised interesting hypothesis related to the fact that what we interpret as bright signal from PLS could actually be related to incompletely suppressed fluid due to field inhomogeneity. We fully agree that this region is very sensitive to B0 and B1 inhomogeneity but it is also important to recall that the new generation of T2 preparation pulse uses adiabatic refocus and are more insensitive to B0/B1 inhomogeneity than older non-adiabatic pulses. Furthermore, the thin band that we highlighted between structures that look like saccule and utricle in the center of the vestibule cannot be explained by such field inhomogeneity artifact at the bone interfaces. Also, the signal from the center of the vestibule that we interpret as ELS is lower than, not only the periphery (i.e.; PLS), but also lower than signal from CSF which is why we believe we capture subtle differences in fluid composition with such T2Prep module and this cannot only be images related to artifacts at the periphery of the vestibule. But, we also agree with Drs Naganawa and Sone that signal is higher on the lateral wall of the vestibule which can indeed be related to field inhomogeneity in this particular region on top of higher signal from PLS. Also, another possibility could be the inhomogeneity of composition in PLS. PLS is affected by local substance-specific homeostatic processes relating to the many cell types contacting it [1]. Variation in protein composition is unknown, but ion composition is known to vary by location [2]. Although there is ongoing entry of CSF through the cochlear aqueduct, it is extremely slow, moving at approximately 30 nL/min [3]. Therefore, any substance in the PLS depends on the rate of local kinetical processes and whether they occur slowly or rapidly. For instance, when K+ in the PLS of the scala tympani was elevated by perfusion, it was eliminated rapidly with a nine-minute half-time [4]. In addition, there is not only liquid in the PLS but also a small amount of nonfluid comportment, including sensories, vessels and connective tissues, which can also affect the signal in the PLS. All these complex factors could affect the signal of PLS.

Finally, Drs Naganawa and Sone pointed subtle differences between the optimized non-contrast FLAIR and the conventional contrast FLAIR in Fig 4d. We believe such differences are not unexpected. First, resolution is not perfectly the same and subtle difference in image orientations can also induce some differences. More importantly, non-injected and injected images rely on totally different principles. While injection captures gadolinium leakage in PLS, our non-injected method is supposed to capture directly differences of composition between ELS and PLS. Especially, we cannot exclude that distribution of gadolinium could be affected by pressure that increases in the hydrops side; a factor that wouldn’t impact the non-injected approach. The interesting and important finding in Fig, 4d is that PLS signal between the saccule and utricle was observed in the healthy side, but not in the hydrops side in the optimized non-contrast FLAIR similarly to standard contrast FLAIR.

To conclude, several points are still to be elucidated with such non contrast approach and new semiology might have to be described. Even if the non-contrast optimized FLAIR is likely not to replace the gadolinium-based gold standard method for all the patients, we believe it might become an easy method of triage to explore a large number of patients with Méière’s disease like symptoms. Direct comparison of this method with the gadolinium method will be needed to validate this statement.