Previous studies have shown that GLOA is a safe and effective procedure [1, 3, 8, 12, 13], but that it also has very variable effects on post-interventional pain score. A retrospective evaluation by Elsner et al. [8] showed that GLOA could initially lead to a 99% reduction in pain but that an increase in pain scores is also possible. These authors reported a pain reduction in 66% of patients in the medium term (up to 7 days), with 21% of patients reporting a decrease of pain outcomes in the long term (a mean of 3 years). Spacek et al. [12] performed an uncontrolled retrospective analysis of 21 patients and reported a beneficial effect of GLOA on the SCG in patients with trigeminal neuralgia. In a different analysis, this same group showed that GLOA on SCG failed in 37% of interventions [13]. Feigl et al. [9] showed that the stylopharyngeal fascia, which can be found in 84% of cases, could lead to a hampered spread of the solution, with the consequence of a decreased effect on the third or even second branch of the trigeminal nerve.

Elsner et al. stated [3] that different patterns of fluid spread in anatomic regions may lead to these inconsistent post-interventional pain outcomes after GLOA on the SCG. In a cadaver study, Feigl et al. [3] evaluated the distribution patterns after GLOA at the SCG according to different volumes injected. In 30 patients, these authors injected 1 ml of contrast (Jopamiro®; Bracco, Milan, Italy) on the left side (group 1) or 2 ml of contrast medium on the right side (group 2); in ten patients (group 3) they injected 5 ml of contrast on the left side in a more sagittal direction, and 5 ml of contrast on the right site in a more lateral direction, as described by Pejic et al. [15]. Computed tomography (CT) scans were taken directly after the intervention, and the spread of contrast medium was evaluated. In the first group, receiving fluid from a needle in a more lateral direction,, the contrast medium was found in the parapharyngeal space in 67% of interventions. In most of these cases, the contrast medium spread from the base of the skull to C2 and reached the dorsal side of the ICA. In 27% of cases, the contrast medium reached the prevertebral space with spreading from the base of the skull to C3. Higher resistance was noted during the application in the prevertebral space. In the second group, 2 ml of contrast was injected in the parapharyngeal space. In 95% of cases, the contrast spread from C2 to C3, frequently reaching further to the dorsal side of the ICA but never reaching the intervertebral foramen. In 6% of cases, the contrast medium was injected in the prevertebral space with a maximum spread up to the C2. In this group also, a higher resistance was noted during the application. In the third group, 5 ml of contrast was injected into the parapharyngeal space. The fluid was observed to spread from the base of the skull to C4 and also reached the infratemporal fossa and the retromandibular fossa. In addition, the contrast medium was found in the transverse and intervertebral foramen. When 5 ml of contrast was injected in the prevertebral space, the fluid was found in the transverse and intervertebral foramen and also reached different muscles of the neck and spread until the C4. When the stylopharyngeal fascia was present, the contrast medium reached various anatomical structures, including the SCG, ICA, internal jugular vein, glossopharyngeal, accessory, vagus and hypoglossal nerves when injected in the parapharyngeal space. When the stylopharyngeal fascia was absent, 2 ml of contrast medium reached the infratemporal fossa. When the liquid was injected into the prevertebral space behind the prevertebral fascia, which acts like a barrier, the parapharyngeal space with the SCG was not reached [3].

Artefactual changes in locations due to postmortem changes may lead to different findings in cadaveric studies compared to in vivo evaluations. Hence, MRI can be applied to evaluate these postmortem findings in living participants.

To the best of our knowledge, our study is the first published report of an MR imaging analysis focused on fluid spread in the area of the SCG immediately after GLOA. On the basis of recent anatomic studies [16], MRI allowed the detection of SCG, with the longus capitis muscle and ICA as landmarks. In our patients, SCG was found at the level of the C2-C4 vertebrae, with a fusiform-elongated shape. Also, different anatomical spaces, in terms of the parapharyngeal space and its “medial compartment,” and prevertebral space, were clearly identified on MRI. The injected fluid was visible in the T1-weighted sequences, as expected.

In this study, we injected 2 ml of fluid transoral into the pharyngeal recessus of Rosenmüller located dorsal to the palatopharyngeal arch. In more than 50% of interventions, the main solution spread reached from C1 to C3, with the furthest craniocaudal solution spread being 63.5 mm. The solution was mostly found in the parapharyngeal space and its “medial compartment,” followed by the “medial compartment” alone. In 93.4% of interventions, the fluid was observed to have spread out completely in the area of the SCG, while in 6.6% of interventions, the solution was found in the prevertebral space and hence did not affect the SCG. In only 13% of cases did the solution come into contact with the dorsal part of the ICA, and no hampered spread due to anatomic distortions was found.

In the cadaveric studies of Feigl et al. [3, 9], 2 ml of contrast medium injected in the parapharyngeal space spread from C2 to C3 and reached the dorsal side of the ICA. Also, in 6% of cases, the solution was found in the prevertebral space. Although these studies described a higher resistance when injecting the fluid in the prevertebral space, we cannot confirm this finding.

We found a significant correlation of solution spread in the parapharyngeal space and its “medial compartment” and pain relief compared to solution spread in the prevertebral spread or the “medial compartment” alone. As described above, the prevertebral fascia is a solid barrier between the prevertebral space and parapharyngeal space, and hence the SCG could not be reached by the solution [3].

The furthest solution spread to the lateral was 21.2 mm, while it was 15.2 mm to the sagittal. In combination with the craniocaudal solution spread, we assumed a cloud-shaped solution spread. To demonstrate an association between distribution patterns and secondary outcomes, we assumed an ellipsoid-shaped volume spread. Interestingly, we found a highly significant correlation of the calculated volume and the solution spread in the parapharyngeal space and its “medial compartment.” We also showed a highly significant correlation between the volume of solution spread and change in NRS.

Feigl et al. [3, 9] reported a regionally confined solution spread when the solution was injected in the parapharyngeal space when a stylopharyngeal fascia was present. We can now confirm these findings of their cadaveric studies as we found the stylopharyngeal fascia in 100% of cases and a regionally confined solution spread when the solution was injected in the parapharyngeal space and its “medial compartment.” As shown in the cadaveric studies, 2 ml of the solution could easily reach the SCG; hence, in a clinical setting, this leads to a decrease in the NRS. The stylopharyngeal fascia is the barrier to the infratemporal and retromandibular fossa. Unfortunately, this stylopharyngeal fascia is inconsistent [10], which may lead to inconstant effects of GLOA at the SCG in patients with trigeminal neuralgia [8, 9, 12, 13].

Furthermore, based on the spread of solution in the area of the SCG, contact of the solution with different anatomical structures and volume of the solution, we estimated a correlation with post-procedural pain scores and side effects. As described above, it was possible to correctly estimate the effects of GLOA at the SCG in 93% (95% CI 68.1–99.8) of cases, which clearly supports the results of cadaveric studies. In 100% of cases, we correctly estimated no minor or major complications.

In two thirds of the interventions reported here, a distinct swelling of the pharyngeal wall was observed after the injection. This led to the hypothesis that the solution may have reached the correct space in front of the prevertebral fascia and that good pain relief should be expected. Unfortunately, no statistical correlation between the swelling and both post-procedural pain scores and volume of solution spread was found. Furthermore, in one intervention, the solution was partially found in the pharyngeal wall and may be a correlated with this post-procedural pharyngeal swelling. No dysphagia was reported in patients with pharyngeal swelling.

Pejic [15] first described the transoral blocking technique of SCG in 1965 in which under direct visual control, the cannula is inserted into the pharyngeal recessus Rosenmüller, in a strictly laterodorsal approach. When the needle is advanced in a more mediodorsal direction, an injection in the prevertebral space is possible. In their cadaveric study, Feigl et al. [3] showed that a more mediodorsal-directed insertion of the needle leads to an increased risk of a possible injection dorsal of the prevertebral fascia, with a resulting failure of the block. In our study, we found no correlation between post-procedural pain scores and different approaches of the needle. Interestingly, a correlation between a strict lateral needle approach and higher volumes of solution spread just fell short of statistical significance. Taken together, these finding lead to the hypothesis that a strict laterodorsal needle direction is preferred to avoid possible block failures; however, further investigations are needed to confirm these findings.

In addition, we found evidence that higher pre-procedural pain intensity is correlated with higher post-procedural pain reduction. It has also been shown that lower post-procedural pain scores increase patient satisfaction. Further effects of GLOA at the SCG on sleep quality, quality of life and patient satisfaction clearly missed reaching statistical significance. Further studies with a larger study population are clearly needed.

There are several important limitations that deserve mention due to the explorative character of this study. First, only limited data are available due to the small study population. While this limitation restricts the number of significant results, the hypothesis cannot be with 100% certainty rejected based on the non-significant results. Secondly, the NRS score of two patients was “0” or “1” before the first intervention, meaning that only slight improvements were possible. These well-known patients of our pain clinic are profiting from a long-term effect of GLOA. As a function of pre-interventional NRS scores, improvements showed an “inclination of distribution.” In general, the statistical methods used take both deviations from the mean and median in both directions into account, which relativizes these findings.

Despite these limitations, we feel that our standardized procedure as well as the separate review of these blinded MRI scans and the correlation with cadaveric studies allow us to make valid clinically relevant conclusions.