Depression, Depressionen

Effect on Function

Table 6 presents the MOS SF-36 overall and subscale results (baseline to exit) for all patients, as well as separateiy for responders and nonresponders (defined by > 50% reductjon in baseline HDRS₂₈ Note that baseline function was remarkably low (e.g., roie emotional = 6.9. roie function = 41.4, vitality = 9.0. social function = 22.0, etc.). Response (by HDRS₂₈ was associated with highiy clinically significant increases in the mental component, role function, vitality, social function, role emotional function. and mental health. Nonresponders did not change (save for a statistically significant increase in social function of 10 points). Of the 12 responders, 25% achieved exit role emotional ratings that equaled or exceeded population norms, indicating that these patients achieved normal functioning.

Which Patients Respond to VNS?

A logistic regression analysis was conducted using HDRS₂₈ response and the following prognostic factors:

HDRS₂₈, total baseline score, diagnosis (major depressive or bipolar disorder), length of index MDE, total iength of illness, age, prior ECT response, VNS stimulation output current, and TST score. A univariate model indicated that the only potentiaily significant factors concerned prior ECT response and VNS stimulation output current (mA); however, the relationship between ECT response and VNS response did not reach statistical significance (p = .10) in a multivariate model that included both ECT response and output current. Lower levels of VNS stimulation output current were associated with better treatment response (p = .07) in the same multivariate model.

Of seven patients not responding at all (either partially or transientiy) to ECT. only one responded to VNS. Within the five categories of ECT response, the logistic regression odds ratio was strongest when patients who had completely failed to respond acutely to ECT were compared with all other patients (i.e., those who never had ECT combined with those with transient or partial responses).

Adverse Events

No patient discontinued the acute study due to adverse events. Reported adverse events were similar to those in previous studies of epilepsy (Table 4). Some adverse events were related to implantation and stimulation, with most of the treatment-related adverse events being stimulation related. Nine patients (30%) reported pain at the incision site, which typically dissipated over 1 to 2 weeks. The most common events reported as possibly, probably, or definitely related to stimulation were hoarseness (60%), throat pain (27%), headache (30%), shortness of breath (23%), general pain (23%), and neck pain (17%). All events had been reported with VNS in previous epilepsy studies (Handforth et al 1998; Vagus Nerve Stimulation Study Group 1995) - although headaches were not as commonly associated with stimulation in the epilepsy trials. In general, stimulation-related adverse events (in cluding hoarseness and throat pain) were mild and well tolerated, and they occurred only when stimulation was on. One patient developed hypomania. which subsided with stimulation reduction.

Additionaily, three patients (10%) reported abnormal wound healing, which involved a longer time for implant incisions to heal. All patients‘ wounds healed without significant intervention. These events all occurred at one site; the surgeon has since modified his incision closure technique. After implantation, two patients (7%) at one site experienced rashes, which eventually subsided.

Six clinically significant adverse events occurred, five during the acute study and one during the long-term follow-up. One event, infection, was related to implantation. Two separate events of leg pain reported by one patient may have been related to implantation. Two events (agitation / panic during the acute study and agitation / irritability /dysphoria in the long-term follow-up) were reported as possibly related to stimulation. One event of worsened depression due to benzodiazepine withdrawal was reported as not related to stimulation.

Safety Testing and Post-Acute Study Follow-Up

Twenty-four-hour Holter monitor results revealed no significant cardiac changes when comparing baseline with acute study exit recordings. Nearly all patients (29/30. 97%) have continued to receive VNS treatment after exiting the acute study. One patient had the NCP generator explanted after 10.8 months of follow-up, due to an inability to sustain antidepressant effects at levels of stimulation that were comfortable. To date, all 10 patients who responded acutely and for whom follow-up data are available have maintained response (Table 7), though two patients experienced a transient worsening on one follow-up occasion. Recall that changes to stimulation parameters and mood-stabilizing medications were both permitted and made during follow-up. At their most recent follow-up visits, seven of these 10 acute study responders demonstrated a complete response (HDRS₂₈< 10).

Discussion

This is the first report of VNS in adult outpatients with severe, nonpsychotic, treatment-resistant MDEs. Response rates of 40% (by HDRS₂₈ and by CGI) or 50% (MADRS), as well as the 17% complete response (remission) rate (exit HDRS < 10), suggest efficacy in this very treatment-resistant population. Responses occurred between 1 and 10 weeks following the initiation of stimulation.

To date, all of the 10 responders for whom we have follow-up visit data after acute study exit have basically sustained the response status over the 4-‐9 months following implantation. Additionally, as of the most recent follow-up visit, seven of these 10 acute-phase responders have attained or remained in remission (HDRS &lt; 10). No significant correlates of VNS response were found, though a larger sample is needed to evaluate both prior ECT response and output current (mA) settings as potential predictors. Studies of VNS in epilepsy have found no relationship between response and output current. Failure to respond at all (i.e., not even partially or transiently) to prior ECT may be a possible predictor of nonresponse to VNS, as might higher stimulation currents. A larger sample is needed.