Anodal frontal tDCS for chronic cluster headache treatment: a proof-of-concept trial targeting the anterior cingulate cortex and searching for nociceptive correlates

Thirty-one patients (9 females) suffering from rCCH were recruited in our headache clinic (University Department of Neurology, CHR Citadelle, Liège, Belgium). Six patients dropped out during the first week of tDCS treatment because of local skin abrasion and/or inefficacy (n = 4), or unrelated health problems (n = 2). Two patients did not perform the treatment. These 8 patients were not included in the efficacy analysis.

All patients suffered from the chronic form of cluster headache (CCH, ICHD 3 beta 3.1.2 [2]) (mean chronic phase duration: 11 ± 9 yrs) and had been refractory to at least 3 adequate preventive treatments [4], including methylprednisolone, verapamil, lithium carbonate, topiramate and suboccipital betametasone-lidocaine infiltrations. At the beginning of the study, 19 out of the 23 patients were under preventive treatment (stable for at least 2 months) and were allowed to continue it throughout the trial. One patient had percutaneous occipital nerve stimulation for 8 years (Tables 1 and 2). To be included in the trial, patients had to provide a 4-week headache baseline paper diary and to suffer at least 4 CH attacks per week. Other inclusion criteria were absence of other significant medical or psychiatric conditions and personal or family history of seizures. The 23 patients who treated themselves for more than a week (mean age: 49 ± 10 yrs.; 3 females; mean disease duration: 14 ± 8 yrs) were included in the intention-to-treat (ITT) analysis. Twenty-one out of them achieved 4 weeks of treatment while two patients dropped out before this time period because of treatment inefficacy. The 10 patients first enrolled among the 21 stopped tDCS treatment after 4 weeks and were followed for 2 weeks afterwards. The 11 following patients continued tDCS for another 4 weeks to complete a total treatment of 8 weeks, except for 1 patient who dropped out due to lack of efficacy. In this sub-group, subsequent follow-up information was available in 6 patients. Thus, 21 patients (mean age 49 ± 10 years) were available for a per-protocol (PP) analysis of a 4-week treatment effect, and 10 patients for a PP analysis of 8 weeks of treatment. The patients’ allocation and disposition are depicted in Fig. 1.

Bipolar transcutaneous tDCS was applied with a novel portable user-friendly battery-driven device developed by Cefaly Technology® (Seraing, Belgium). The first 4 patients were provided with sticking electrodes containing a special conductive gel (Spes Medica®, Genova, Italy – anode: 35 × 45 mm, cathode: 40 × 90 mm), but developed a transient electro-chemical skin irritation under the cathode after a few days, therefore the treatment was immediately discontinued in these patients. In subsequent patients we employed sponge-electrodes (80 × 60 mm, Spes Medica®, France), moistened with saline, we had used previously in a migraine study with a non-portable tDCS device [23]. The anode was fixed with elastic straps (width 100 mm, Spes Medica®, France) over Fz (10–20 system), the cathode over the spinous process of C7 (Fig. 2). No local skin irritation was seen with sponge electrodes.

All patients were trained to adequately position the electrodes and use the device before starting the trial. Stimulation intensity was set at 2 mA, and patients were asked to apply tDCS outside an attack as a preventive treatment, once daily during 20 min where after the device switched off automatically. The stimulation parameters were set in accordance with safety recommendations [24, 33]. Adherence to the tDCS treatment was monitored with an in-built software designed by Cefaly Technology®. During the trial patients could treat their CH attacks as usual, the majority of them using injectable sumatriptan and oxygen inhalation.

Simulations of absolute value of electric field intensity and electric potential distributions were performed using COMETS [34], a MATLAB (The MathWorks Inc.) toolbox for simulation of local electric fields generated by tDCS, based on the electrostatic finite element method (FEM). Parameters of tDCS (electrode size and placement as well as current intensity) introduced in the model were those applied to patients (described in detail above). Simulation results were imported in Tecplot® (Tecplot Inc., WA, US) for 3D visualization (Fig. 2).

The patients filled in cluster headache paper diaries at least one month before beginning the trial (baseline), during the whole tDCS 4- or 8-week therapy and at least 2 weeks after the end of the treatment. Attack occurrence, intensity (rated 1-mild to 4-worst), duration (minutes) and use of acute treatment (injectable sumatriptan, oxygen inhalation, analgesics) were recorded.

There are unfortunately no clinical biomarkers of ACC activation. Searching for changes in frontal functions associated with tDCS, we determined in all patients a Frontal Assessment Battery (FAB) score before and after treatment [35]. The FAB consists of six subtests exploring conceptualization, mental flexibility, motor programming, sensitivity to interference, inhibitory control and environmental autonomy [35].

Depression scores were also determined before and after tDCS with Beck’s Depression Inventory (BDI) [36].

Patients were interrogated about possible side effects of tDCS at each visit and asked to immediately inform the Headache Research Unit team in case of any adverse event.

Eighteen out of 21 patients accepted to undergo thermal quantitative sensory testing (QST) and 11 to have nociception-specific blink reflex (nBR) recordings before and after treatment.

During QST, using a thermode (Advanced Thermal Stimulator-Medoc™ USA), we determined sensory and pain thresholds to cold (CST and CPT) or warm stimuli (WST and WPT) bilaterally over the forehead and the volar side of the wrist. The device allows to deliver stimuli between − 10 °C and + 54 °C. The thresholds were determined in steps of 1 °C/second starting at 32 °C. The subjects were instructed to press a button when they perceived the stimulus and when it became painful. The mean of three successive measures was taken as threshold value for each variable.

Nociception-specific blink reflexes (nsBR) were recorded as previously described [37]. Briefly, surface recording electrodes were placed bilaterally over orbicularis oculi muscles, and electrical stimulation was performed supraorbitally with a concentric electrode (central cathode: 1 mm; insert: 8 mm; anode: 23 mm). Monopolar square pulses of 0.2 ms duration were delivered at a pseudo randomized interstimulus interval between 15 and 17 s. We first determined electrical sensory and pain thresholds using ascending and descending steps of 0.2 mA intensity (Digitimer stimulator DS7A). To elicit the nsBR, the final stimulus intensity was set at 1.5 times the individual pain threshold. Sixteen rectified electromyographic responses were recorded and averaged off-line (CED 1401 and 1902 devices, Signal 4.11 Software, Cambridge Electronic Design, Cambridge, UK). The first response of each session was discarded to avoid contamination with startle responses. The remaining 15 sweeps were averaged in three sequential blocks of five responses. The amplitude of the R2 response was calculated for each block and expressed as area under the curve (AUC). Results were normalized using R2 AUC divided by the square of stimulus intensity (AUC/i²). Habituation of the nsBR was calculated as the percentage change of the R2 AUC between the 3rd and the 1st block of averages and also expressed as the regression slope of the R2 AUC over the three successive blocks of five responses.

The primary outcome measure was the change of weekly CH attack frequency during and following tDCS treatment, compared to the mean weekly frequency during the 4-week baseline. Secondary outcome measures were change in attack intensity and duration, and acute medication use.

As mentioned above, 8 patients were not included in the analysis because they applied tDCS for less than 1 week. Two patients stopped treatment before the 4-week term and were considered protocol violators; their data were handled on a “last value carried forward” basis for the ITT analysis. Twenty-three patients were thus available for intention-to-treat (ITT), 21 for per-protocol (PP) analysis of the effects of daily tDCS treatment during 4 weeks. A subgroup of 10 patients was available for assessing the effect of an 8-week treatment.

Like in a study on neuropathic pain [38], QST data were first standardized and then entered in a non-hierarchical K-means cluster analysis. This analysis was employed in order to identify subgroups of patients with distinct sensory profiles and their possible correlation with treatment outcome. We searched if the tDCS treatment effect was correlated with pre-treatment pain thresholds using Pearson’s correlation analysis and if tDCS had an effect on pain thresholds with mixed-design ANOVA.

The changes in outcome measures in the per-protocol (PP) analysis (N = 21) over 4 weeks of treatment are graphically depicted in Fig. 3. Mean weekly attack frequency decreased significantly from 15.33 ± 13.12 at baseline to 9.91 ± 11.72 after 4 weeks of tDCS (− 5.43/35%, p < 0.001). The 50% responder rate was 38%. Mean attack duration decreased from 47.70 ± 50.55 min at baseline to 32.62 ± 28.38 min (p = 0.020) and mean attack intensity from 3.2 to 2.5 (p = 0.016). Weekly use of abortive treatments decreased from 13.82 ± 13.83 at baseline to 8.00 ± 8.81 at 4 weeks (p = 0.006) (Fig. 3).

Favourable outcomes were sustained over time (Friedman test p = 0.049) (Fig. 3). In the subgroup of patients who treated themselves with tDCS for 8 weeks (N = 10), weekly CH attack frequency decreased from 18.90 ± 16.01 at baseline to 12.30 ± 16.57 (p = 0.041, Fig. 4). The 50% responder rate was 50%. Reductions in mean attack duration, severity and acute treatment use did not reach the statistical level of significance in this smaller subset of patients.

In the intention-to-treat analysis of all patients who performed at least 1 week of tDCS (N = 23) the results were similar showing a significant decrease of attack frequency after 4 weeks of treatment (p < 0.001).

A pooled analysis of compliance revealed that patients who completed the protocol (4 or 8 weeks) had used the tDCS device 87% of the recommended time.

Overall, thermal QST results were not modified by tDCS whatever modality (cold/warmth), threshold (sensory/painful), side (right/left) or stimulus location (forehead/wrist) was considered (all p > 0.1). Along the same line, electrical thresholds and nsBR results were not modified by tDCS (all p > 0.1).

Searching for correlations between treatment response and baseline thermal pain thresholds, we found that patients who perceived pain at more extreme temperatures exhibited a better response to tDCS. Individual baseline cold pain threshold (CPT) correlated with the percentage reduction of attack frequency after 4 weeks of tDCS (N = 21, r = 0.45, p = 0.042) and concordantly, heat pain threshold (HPT) anti-correlated with tDCS-induced attack frequency reduction (r = − 0.45, p = 0.041, Fig 5). A data-driven K-means cluster analysis revealed 2 distinct QST profiles: patients with low (‘hypersensitive’, N = 9) and patients with high pain thresholds (‘hyposensitive’, N = 12). The tDCS-induced reduction in CH attack frequency was greater in ‘hyposensitive’ (− 6.67 attacks/week, p = 0.014) than in ‘hypersensitive’ patients (− 3.78 attacks/week, p = 0.049).

The mean frontal assessment battery (FAB) score significantly increased after tDCS, (from 16.58 ± 1.46 to 17.16 ± 1.17, N = 19, p = 0.01). There were no significant changes in BDI scores (13.18 ± 18.88 before vs 12.41 ± 9.22 after tDCS, p > 0.1).

The sponge electrodes were well tolerated and did not produce any skin abrasion, like in our previous tDCS study in migraine [23]. Besides a slight and transient tingling sensation at the electrode site, frequently reported with tDCS [24], there were no treatment-related adverse effects. Among the 8 patients who stopped tDCS during the 1st week, 2 had actually not switched on the device at all while the 4 others applied tDCS only for a few days because of electrochemical skin irritation related to the use of sticking electrodes. These electrodes had been tested with the tDCS device by the manufacturer before the study. We hypothesize that the repetition of tDCS could be responsible for this skin irritation. Conversely, sponge electrodes were very well tolerated at long–term. Two patients dropped out for unrelated health problems: ENT cancer and peritonitis.