Chronic migraine (CM) is a disabling neurological disorder affecting about 2% of the population [1]. It occurs on 15 or more days per month for more than 3 months, having the features of migraine headache on at least 8 days per month. People with CM have impaired quality of life, more severe disability, and greater economic burden than patients with episodic migraine (EM). Furthermore, patients with CM are often poor responders to prophylactic treatments.

Recently, onabotulinumtoxinA (OBT-A) was proposed as a preventive treatment option for CM. Its efficacy for the CM prophylaxis has been demonstrated in two well-designed phase III clinical trials, PREEMPT 1 and 2 trials [2, 3]. A pooled analysis of these trials showed that OBT-A was significantly more effective than placebo in reducing the mean frequency of days with headache and headache episodes.

The use of OBT-A for the prophylaxis of CM has also been endorsed by several international societies, usually as second-line treatment. In Italy, onabotulinumtoxinA was approved for the treatment of CM in 2013 in patients who have failed, or do not tolerate, oral prophylactic treatments. Recent studies demonstrated that the treatment is effective and safe for CM, even if complicated with medication overuse headache (MOH), in the short and long term.

However, the management of OBT-A in clinical practice remains to be fully defined. In particular, it remains to be clarified how long the treatment has to be continued either if clinically effective or not, and, if effective, whether and when it has to be discontinued and for how long. Current literature have not addressed these and other questions and currently different centers may administer OBT-A in different ways.

The aim of the present study was to answer these questions from the real-life clinical practice by three third level Italian headache centers in order to help finding the best management of OBT-A in CM patients.

First, we compared data from two different centers (Roma, Campus Bio-Medico and Milano, Istituto Neurologico Besta). Both centers performed treatment with OBT-A in patients with CM, following the PREEMPT protocol, for at least four injections, with a 12-week span.

We retrospectively reviewed medical records of consecutive patients treated in these centers from November 2013 to November 2017. Our analysis proceeded in order to answer three questions: (a) if there is consistency in patient selection between the centers, (b) if there is consistency in the treatment efficacy between centers, and (c) if, halfway during the treatment, it is possible to identify a subset of patients who will not benefit from the continuation of the treatment. For this latter purpose, we defined as responders patients experiencing 30% or more reduction in headache frequency at an intermediate time point (3 months after the second injection).

We then compared both centers with a third one (Bologna, IRCCS Istituto delle Scienze Neurologiche, Ospedale Bellaria). This center performed the PREEMPT protocol evaluating patients at 6 months, and continued the treatment only in responder patients (patients with 30% or more reduction in headache frequency or a significant reduction in disability).

Set time points were T0 (baseline, before the first injection), T2 (6 months, 3 months after the second injection and just before the third injection), T4 (1 year, 3 months after the fourth—and last—injection).

We found 53 patients in Roma and 62 in Milano. Three patients in Roma and 3 patients in Milano interrupted the treatment before the third injection, and the other 5 patients in Milano interrupted between the second and the fourth injections.

The rate of missing data was 6.8%; we chose to deal with them by using a multiple imputation.

Regarding patient selection, we found a remarkable consistency between centers. The typical CM patient selected for OBT-A injection was female (Roma 83%, Milano 83%, p = .915), in her fifties (mean age, SD Roma 51.6 (10.3), Milano 51 (8.2), p = .740). Monthly headache frequency (mean, SD Roma 21.17 (6.66), Milano 20.15 (6.22), p = .396), symptomatic drug consumption (mean, SEM Roma 20.23 (0.07), Milano 17.1 (0.01), p = .160), HIT6 score (mean, SEM Roma 65.79 (1.01), Milano 64.83 (0.86), p = .469), and MIDAS score (mean, SEM Roma 58.54 (0.40), Milano 64.31 (0.24), p = .646) did not differ in the two centers.

The above variables were comparable also at different time points. At each time point, indeed, non-significant differences were found for each variable between centers (except for HIT6 at T2, but without a significant difference at T4).

We found a significant benefit from the treatment at T2. At this intermediate stage, all the variables in each center improved with statistical significance (except for HIT6 score in Rome). We found a mean T0-T2 decrease in monthly headache day frequency of 6.5 days in Roma (p = .002) and 7.5 days in Milano (p ≤ .0001); a mean decrease in monthly symptomatic drug consumption of 5.5 drugs in Roma (p ≤ .0001) and 5 drugs in Milano (p ≤ .0001); a mean HIT6 score reduction of 1.22 in Roma (p = .410) and 4.24 in Milano (p = .001); and a mean reduction in MIDAS score of 26.3 in Roma (p = .020) and of 23.22 in Milano (p = .018).

Results obtained at T4 for each variable did not differ from those at T2, with no statistical significant difference also between centers.

Then, we compared patient with < 30 and ≥ 30% reduction in headache frequency between T0 and T2 (respectively, “non-responder” and “responder”), evaluating if this intermediate cut-off could discriminate which subjects will benefit from continuing the treatment.

The rate of responders presented a trend toward difference between centers (Roma 47%, Milano 64%, p = .072). Nevertheless, responders and non-Responders from the two centers presented similar efficacy trends throughout the treatment. While responders showed a similar course to the one of the global sample (with a substantial improvement between T0 and T2 and stability between T2 and T4), non-responders present little, if any, improvement in headache frequency and drug consumption at any time. In both centers, responders and non-responders presented similar headache frequency and drug consumption at baseline (except for drugs consumption in responders, lower in Roma than in Milano), but the differences became statistically significant at T2 and T4. On the other hand, HIT6 score and MIDAS score, in both centers, presented a similar trend between responder and non-responder, without statistical differences at any time points.

Finally, we tried to compare our results with those in Bologna center with a different longitudinal application of the protocol (exclusion of non-responding patients at T2). A direct comparison between the centers of Roma, Milano, and Bologna was not possible, because of significant difference in the number (n = 128) and in the baseline characteristics of Bologna patients. The Bologna center presented a slightly higher percentage of male patients (male 24%), and a slightly higher, but significantly different, mean headache frequency. This is in part due to the different role of Bologna center due to differences in regional public health system. Another limitation was the lack of homogeneity in data collection across centers, due to the retrospective nature of the study.

After a multiple imputation, a partial analysis of the data from Bologna revealed that patients not excluded from continuing the treatment at 6 months (i.e., patients with ≥ 30% reduction in headache frequency between T0 and T2; 72% of the whole Bologna population) presented a longitudinal trend similar to the responder patients in Roma and Milano, with a constant improvement that was statistically significant only between T0 and T2.

Our analysis confirms that OBT-A treatment following the PREEMPT protocol is an effective treatment for CM. Real-life results present consistency with the literature. Patient selection results are similar to data from PREEMPT trial. Patients evaluation halfway during the treatment could be useful to discriminate ones who will benefit from continuing the treatment from those will not. A more thoroughly analysis is needed to better clarify this question.