Reduced GABAergic cortical inhibition in aging and depression

Abstract

The neurobiology underlying depression in older adults is less extensively evaluated than in younger adults, despite the putative influence of aging on depression neuropathology. Studies using transcranial magnetic stimulation (TMS), a neurophysiological tool capable of probing inhibitory and excitatory cortical neurotransmission, have identified dysfunctional GABAergic inhibitory activity in younger adults with depression. However, GABAergic and glutamatergic cortical neurotransmission have not yet been studied in late-life depression (LLD). Here, we used single- and paired-pulse TMS to measure cortical inhibition and excitation in 92 LLD patients and 41 age-matched healthy controls. To differentiate the influence of age and depression, we also compared these TMS indices to those of 30 younger depressed adults and 30 age- and sex-matched younger healthy adults. LLD patients, older healthy adults, and younger depressed adults demonstrated significantly lower GABA_A receptor-mediated cortical inhibition than younger healthy controls. By contrast, no significant differences in cortical inhibition were observed between older adults with and without depression. No significant differences in GABA_B receptor-mediated inhibition or cortical excitation were found between the groups. Altogether, these findings suggest that reduced cortical inhibition may be associated with both advancing age and depression, which (i) supports the model of depression as a disease of accelerated aging, and (ii) prompts future investigation into diminished GABAergic neurotransmission in late-life as a biological predisposing factor to the development of depression. Given that cortical neurophysiology was similar in depressed and healthy older adults, future prospective studies need to establish the relative influence of age and depression on cortical inhibition deficits.

Introduction

Depression, whether it occurs early in life and persists for decades or develops for the first time in late life, is one of the most common neuropsychiatric disorders in older adults and is associated with increased rates of suicide [1], premature mortality from comorbid illness [2], and increased health care costs [3]. Although standard treatments are effective for some older adults with depression [4], in general, the prognosis for older adults with early- or late-onset depression is poor [5]. Whereas the majority of research on depression to date has focused on younger adults, there is evidence that the molecular correlates of normal aging resemble those observed in depression [6], and it remains unclear how molecular aging may influence the pathophysiology of late-life depression (LLD), particularly with late-onset [7]. For example, overlapping changes in gene expression [8] and reductions in cortical volume [9, 10] occur in both aging and depression. Going forward, the development of improved treatments for LLD may benefit from a better understanding of the pathophysiological similarities and differences of depression in younger and older adults.

Although all currently approved antidepressants target the monoamine system, studies in younger adults with depression have implicated dysregulation of the two main inhibitory and excitatory neurotransmitter systems: gamma-aminobutyric-acid (GABA) and glutamate [11, 12]. Early studies found abnormal peripheral levels of GABA and glutamate in the cerebrospinal fluid or serum of young adults with depression [13, 14]. Magnetic resonance spectroscopy (MRS) studies have since found mostly downregulated cortical GABA concentrations and abnormal cortical glutamate concentrations in patients under age 60 with depression [15]. Similarly, post-mortem studies of patients with depression have reported reductions in the size and density of cortical GABAergic interneurons [16], downregulation of GABA-related genes [17], and altered glutamate N-methyl-D-aspartate (NMDA) receptor levels [18]. Far fewer studies of depression have investigated GABAergic and glutamatergic systems in older adults. One study of peripheral metabolites in older depressed patients found reduced plasma GABA concentrations in patients compared to age-matched controls [19]. Additionally, a post-mortem study of depressed patients aged 30–86 years found a negative correlation between age and the density and soma size of prefrontal cortex GABAergic neurons [16]. In order to inform treatment development for older adults with depression, a clearer understanding of the role of GABAergic and glutamatergic neurotransmission in LLD is needed.

Transcranial magnetic stimulation (TMS) is a non-invasive brain stimulation method that can be used in conjunction with electromyography (EMG) to reliably probe inhibitory and excitatory neurotransmission in the motor cortex. Whereas MRS is believed to measure tonic inhibitory tone [20], TMS paradigms alone can measure the phasic activation of GABAergic interneurons, thus offering unique insight into the synaptic activity of inhibitory and excitatory systems in humans [21]. For example, short-interval cortical inhibition (SICI) and cortical silent period (CSP) TMS measures can be used to investigate cortical inhibitory neurotransmission [22, 23], with good test–retest reliability [24, 25]. SICI measures suppression of the motor-evoked potential (MEP) amplitude when a subthreshold TMS pulse precedes a suprathreshold TMS pulse by 1–6 ms. The CSP measures the duration of EMG activity suppression following a suprathreshold TMS pulse delivered during muscle contraction.

SICI and CSP indirectly measure GABA_A and GABA_B receptor-mediated inhibition, respectively. For example, benzodiazepines act as positive allosteric modulators at GABA_A receptors and reliably facilitate SICI [26], whereas baclofen, a GABA_B receptor agonist, prolongs CSP duration [27]. Likewise, the time course of inhibitory effects observed in SICI are consistent with the fast time course of GABA_A-mediated inhibitory postsynaptic potentials (IPSPs) [28], and the CSP duration corresponds to the peak of slow GABA_B receptor-mediated IPSPs [29]. SICI and CSP durations have both been found to be low in younger patients with depression and treatment-resistant depression [30–32], reflecting impaired GABA_A and GABA_B receptor-mediated inhibition. The association between these TMS measures of GABAergic neurotransmission and LLD has yet to be examined.

TMS can also be used to assess cortical excitation through resting motor threshold (RMT) and intracortical facilitation (ICF) paradigms. RMT is thought to index general neuronal membrane excitability [21], whereas pharmacological studies suggest that ICF reflects the activity of excitatory NMDA receptor-dependent glutamatergic circuits. For example, administration of NMDA receptor antagonists has been shown to decrease ICF [26]. Although RMT and ICF have typically been reported to be normal in younger adults with depression, differences in cortical excitability have been found between younger and older healthy adults [33, 34]. As with TMS measures of cortical inhibition, these TMS indices of cortical excitability have yet to be assessed in older adults with depression.

Here, we conducted the first study of TMS measures of cortical inhibition (SICI and CSP) and excitation (RMT and ICF) in a large sample of older adults with early- or late-onset depression. We hypothesized that older adults with depression would display abnormal GABAergic cortical inhibition and glutamatergic cortical excitation compared to age-matched healthy controls. To dissect the influence of age and depression on these TMS measures, we also conducted exploratory comparisons with younger patients with depression, and younger healthy controls.

Materials and methods

Study sample

Older adults with depression who had enrolled in one of two clinical treatment trials at the Centre for Addiction and Mental Health (CAMH) were invited to participate in the study prior to initiating treatment: 92 older adults with depression participated and 41 age-matched healthy controls were recruited from registries and advertisements. For comparison, we also analyzed published data from 30 younger patients with depression [32], and unpublished data from 30 age-matched younger healthy controls.

Older adults with depression were ≥ 60 years of age, met diagnostic criteria for major depressive disorder (MDD), single or recurrent and with early- or late-onset, as diagnosed by the Structured Clinical Interview for the DSM-IV (SCID-IV), and were free of any current comorbid Axis I disorders, with the exception of anxiety disorders. They had a Montgomery-Asberg Depression Rating Scale (MADRS) [35] score ≥ 15 and a Mini-Mental State Examination (MMSE) [36] score > 24. Patients were excluded if they were taking an anticonvulsant. Benzodiazepine use was permitted, however the dose could not exceed lorazepam at 2 mg/d or equivalent. Zopiclone up to 15 mg/d and trazodone up to 50 mg/d were also allowed. Some patients who were being cross-titrated from another antidepressant were on a low dose of this antidepressant at the time of TMS testing. Treatment resistance was assessed in all patients using the Antidepressant Treatment History Form (ATHF) [37].

All healthy controls were free of any psychiatric disorders based on the Mini International Neuropsychiatric Interview Version 6.0 (MINI 6.0), and screened negative for any substance use on a urine toxicology screen (MEDTOX® Diagnostics, Inc., Burlington, NC, USA).

The Research Ethics Board at CAMH approved the study procedures in accordance with the Declaration of Helsinki. All participants provided written, informed consent.

TMS-EMG procedures

For EMG recordings, disposable 9 mm surface electrodes were positioned over the right abductor pollicis brevis (APB) muscle belly (active electrode) and the interphalangeal joint of the thumb (reference electrode). The ground electrode was positioned at the proximal end of the right forearm at the inside of the elbow joint. Participants were asked to relax their right hand throughout the experiment, and EMG recordings were monitored in real time to verify the relaxed state of the APB muscle. The EMG signal was amplified, filtered (band pass 2–2.5 kHz) and digitized at 5 kHz.

A figure-of-eight coil with 70 mm diameter coil loops and two Magstim 200 stimulators (Magstim, Whitland, Dyfed, Wales), connected by a Bistim module, were used to administer monophasic TMS pulses. The TMS coil was positioned tangentially on the head, over the left motor cortex, at the optimal location for evoking motor-evoked potentials (MEPs) from the APB muscle at rest. The axis of the coil was held at ~ 45° angle lateral to the midsagittal line, in order to induce a posterior-anterior current flow, optimal for trans-synaptic activation of corticospinal neurons [38]. All TMS methodologies adhered to a recently developed checklist for the collection and reporting of TMS data [39].

RMT was defined as the minimum TMS intensity required to produce a ≥ 50 µV peak-to-peak MEP amplitude in the relaxed APB muscle in at least 5 of 10 consecutive trials. The TMS intensity required to evoke MEPs with a peak-to-peak amplitude of ~1 mV (0.5–1.5 mV) in the relaxed APB muscle was then established for use as the test stimulus (TS) intensity in the paired-pulse paradigms [40].

For SICI and ICF paired-pulse TMS measures, a subthreshold conditioning stimulus (CS) at 80% RMT preceded a suprathreshold TS (intensity required to elicit ~1 mV peak-to-peak MEP amplitude) to the motor cortex with an interstimulus interval (ISI) of 2 ms (SICI) or 10 ms (ICF) [22]. For each participant, 36 trials were conducted, 12 for each condition: (i) TS alone (unconditioned response), (ii) CS preceding TS with 2 ms ISI (SICI conditioned response), and (iii) CS preceding TS with 10 ms ISI (ICF conditioned response). ISIs of 4 ms, 15 ms and 20 ms were also studied in a subset of older adults with depression (n = 40), however ISIs of 2 ms and 10 ms were found to elicit the greatest mean inhibition and facilitation, respectively, and are therefore the focus of this paper. Analyses of the SICI and ICF measures with ISIs of 4 ms, 15 ms and 20 ms are described in the Supplementary Materials and Methods. The conditioned MEP amplitude for SICI and ICF is expressed as a ratio of the conditioned response to the unconditioned response, i.e., SICI or ICF ratio = MEP_conditioned/MEP_{unconditioned} [41]. Thus, larger SICI ratios reflect larger conditioned MEPs and lower cortical inhibition.

For the single-pulse CSP paradigm, participants maintained voluntary muscle contraction at 20% of maximum force on a gauge meter using a pinch grip (thumb and index finger). During muscle contraction, 10 single suprathreshold TMS pulses (140% RMT) were delivered to the left motor cortex [23]. An average of the 10 trials was computed off-line, and CSP duration was assessed using the averaged recording [42]. Following a masking procedure, the CSP durations were measured from the onset of the MEP to the return of EMG activity by an experienced rater. Masking was performed using a Matlab script that shuffled and renamed all CSP files in a pseudorandom order. CSP durations were calculated for the shuffled files, which contained no clinical or demographic group indicators. After all CSP durations were extracted, the data were unshuffled. For all TMS paradigms, individual trials were manually checked and excluded if there was excessive noise or artifacts.

Statistical analysis

Demographic characteristics were compared among the groups using a one-way analysis of variance (ANOVA) followed by independent t tests, or using a Chi-Square (χ²) test for categorical variables. The significance level for all analyses was set to an α level of 0.05 (two-tailed). All data are reported as mean ± standard deviation (SD) unless otherwise stated.

Using histograms, boxplots, Q-Q plots, and Shapiro–Wilk test, we determined that RMT and the SICI and ICF ratios were non-normally distributed. RMT and SICI data were log-transformed, and ICF data square-root-transformed, to attain normal distributions. For RMT, CSP, SICI, and ICF measures, the primary analysis consisted of a two-way ANOVA with age (two levels: older and younger) and diagnosis (two levels: depression and no depression) as the between-subjects factors. Secondary analyses consisted of planned pairwise comparisons between subgroups, uncorrected for multiple comparisons. In addition, sensitivity analyses were performed to examine the robustness of the findings in the presence/absence of factors that may influence the TMS measures of GABAergic/glutamatergic functioning: antidepressant or sedative use, treatment resistance, late (≥ 60) or early (< 60) onset of depressive symptoms, and comorbid anxiety disorders (see Supplementary Materials and Methods).

Depending on the distributions of the measures of interest, exploratory Pearson or Spearman correlation analyses were performed. Correlation analyses were performed between each TMS measure and age across and within all groups, and between TMS measures of cortical inhibition and excitation across and within groups. In the LLD subgroup separately, each TMS measure was correlated with MADRS score and age of onset of the first depressive episode. in addition, greater medical burden is closely linked to both old age and depression [43, 44], and may mediate any associations between neurophysiological measures and LLD. Therefore, in the LLD subgroup, each TMS measure was also correlated with the Cumulative Illness Rating Scale for Geriatrics (CIRS-G) score [45].

Results

See Table 1 for a summary of demographic, clinical, and neurophysiological information across groups. Younger healthy controls reported significantly more years of education than older adults with depression (t₁₁₉ = 2.9, p = 0.005).

Table 1.

Patient demographic, clinical, and neurophysiological characteristics

	Older adults with depression (n = 92)	Older healthy adults (n = 41)	Younger adults with depression (n = 30)	Younger healthy adults (n = 30)	Statistic	p value
Demographic characteristics
Age (years)	66.8 ± 6.1	69.0 ± 8.3	44.8 ± 10.5	44.9 ± 10.8	F3,189 = 105.7	<0.001
Sex	62 F/30 M	23 F/18 M	20 F/10 M	18 F/12 M	χ2(3) = 1.9	0.60
Education (years)	14.5 ± 2.9	15.4 ± 2.3	NA	16.2 ± 2.9	F2,158 = 4.9	0.009
Handedness	66 R/8 La	40 R/1 L	30 R/0 L	28 R/2 L	χ2(6) = 11.3	0.08
Clinical characteristics
MADRS score	25.7 ± 5.4	NA	NA	NA	−	−
HAM-D score	NA	−	25.6 ± 3.0	−	−	−
MMSE score	29.0 ± 1.1	NA	NA	NA	−	−
CIRS-G score	6.5 ± 3.7	NA	NA	NA	−	−
Late-onset (≥ 60 years), no. (%)	19 (21%)	−	−	−	−	−
Treatment-resistant, no. (%)b	69 (75%)	−	19 (63%)	−	χ2(1) = 1.5	0.22
Comorbid anxiety disorder, no. (%)	32 (35%)	−	14 (47%)	−	χ2(1) = 1.4	0.24
Medication use, no. (%)c
Antidepressants	45 (49%)	−	18 (60%)	−	χ2(1) = 1.1	0.29
Benzodiazepines	27 (29%)	−	11 (37%)	−	χ2(1) = 0.6	0.45
Neurophysiological characteristics
TS intensity (%)	63.5 ± 15.3	63.7 ± 12.7	NA	59.5 ± 12.7	F2,159 = 1.0	0.38
TS mean MEP amplitude (mV)	0.98 ± 0.47	0.86 ± 0.34	0.95 ± 0.24	0.93 ± 0.30	F3,185 = 0.9	0.45

Abbreviations: R = right, L = left, MADRS = Montgomery-Asberg Depression Rating Scale, MMSE = Mini-Mental State Examination, HAM-D = Hamilton Rating Scale for Depression, RMT = resting motor threshold (% maximum stimulator output), NA = not available, TS = test stimulus, TS intensity = intensity required to elicit 1 mV peak-to-peak MEP amplitude (% maximum stimulator output), MEP = motor-evoked potential

^a Handedness data missing for 18 patients

^b Treatment resistance defined as non-response to an adequate antidepressant trial (a score ≥ 3 on the Antidepressant Treatment History Form)

^c Medication use at the time of neurophysiological testing

Cortical inhibition

SICI and ICF data were missing for four younger adults with depression, thus 26 younger adults with depression were included in the SICI and ICF analyses. For SICI, significant main effects of age and diagnosis were observed (see Table 2). The interaction between age and diagnosis was not significant (F_1,185 = 2.73, p = 0.10). Compared with younger healthy controls, SICI was significantly reduced in older adults with depression (t₁₂₀ = 3.58, p < 0.001, d_Cohen = 0.75), older healthy controls (t₆₉ = 2.45, p = 0.017, d_Cohen = 0.59), and younger adults with depression (t₅₄ = 2.39, p = 0.02, d_Cohen = 0.64). SICI did not differ significantly between older adults with depression and older healthy controls (t₁₃₁ = 0.63, p = 0.53) or between older and younger adults with depression (t₁₁₆ = 0.29, p = 0.78). See Fig. 1 for an illustration of SICI across groups.

Table 2.

TMS measures of cortical inhibition and excitation by group

	Older adults with depression (n = 92)	Older healthy controls (n = 41)	Younger adults with depression (n = 30)	Younger healthy controls (n = 30)	Statistic	p value
Short-interval cortical inhibition ratio	0.64 ± 0.49 (L-O: 0.58 ± 0.20) (E-O: 0.66 ± 0.54)	0.60 ± 0.39	0.63 ± 0.41	0.37 ± 0.25	Age: F1,185 = 4.14 DX: F1,185 = 5.61	0.043 0.019
Intracortical facilitation ratio	1.83 ± 0.84 (L-O: 1.90 ± 1.04) (E-O: 1.81 ± 0.79)	1.77 ± 0.59	1.52 ± 0.54	1.64 ± 0.72	Age: F1,185 = 3.63 DX: F1,185 = 0.04	0.058 0.84
Cortical silent period duration (ms)	124.8 ± 33.1 (L-O: 111.4 ± 29.2) (E-O: 128.3 ± 33.3)	129.4 ± 29.4	127.2 ± 33.4	135.3 ± 34.1	Age: F1,187 = 0.62 DX: F1,187 = 1.48	0.43 0.23
Resting motor threshold (%)	49.1 ± 10.5 (L-O: 46.05 ± 7.8) (E-O: 49.9 ± 11.1)	48.8 ± 7.9	47.6 ± 16.6	48.0 ± 7.5	Age: F1,189 = 1.22 DX: F1,189 = 0.004	0.27 0.49

L-O = Subgroup of older adults with late-onset depression (≥ 60 years, n = 19); E-O = Subgroup of older adults with early-onset depression (< 60 years, n = 73). Resting motor threshold = % maximum stimulator output; DX = Diagnosis of depression. Subgroup data not included in ANOVA statistics in right-most columns

Fig. 1.

Mean % cortical inhibition in the SICI paradigm by group. % SICI was calculated as (1–SICI ratio) × 100%. Hence, a higher SICI ratio reflects poorer inhibition, whereas a higher % SICI reflects greater inhibition. Error bars represent standard error of the mean. MDD = major depressive disorder patients, HC = healthy controls

Findings of lower SICI in older adults with depression compared with younger healthy controls remained significant in the sensitivity analyses that examined the influence of antidepressant or benzodiazepine/sedative use, treatment resistance, late or early onset of depressive symptoms, and anxiety disorder comorbidity (see Supplementary Materials and Methods).

CSP data were missing for one older adult with depression and one younger adult with depression, thus 91 older patients and 29 younger patients with depression were included in the CSP analyses. For CSP, we did not observe significant main effects of age or diagnosis, or a significant age × diagnosis interaction (F_1,187 = 0.11, p = 0.74). See Table 2 for mean SICI ratios and CSP durations across groups.

Cortical excitation

See Table 2 for mean RMT and ICF ratios across groups. No significant age or diagnosis main effects, and no significant age × diagnosis interaction (F_1,189 = 0.46, p = 0.50), were observed for RMT. Similarly, no significant main effects or interactions were found for ICF (age × diagnosis interaction: F_1,185 = 0.20, p = 0.66).

Correlations with TMS measures

Across all groups, a weak positive correlation was observed between age and the SICI ratio (Spearman’s rho (r_s) = 0.17, p = 0.019), reflecting an association between diminished cortical inhibition and advancing age. The correlation between age and SICI remained significant in the subgroup of healthy controls (r_s = 0.29, p = 0.016), but not in those with depression (r_s = 0.07, p = 0.47). The CIRS-G score, which reflects chronic medical illness burden, was also positively correlated with the SICI ratio in the subgroup of older adults with depression (r_s = 0.26, p = 0.014). No other significant correlations with the TMS measures were observed.

Exploration of the relationships between TMS measures also found no significant correlations between SICI and ICF (r_s = 0.097), SICI and CSP (r_s = −0.092), or ICF and CSP (r_s = 0.007) across groups. In only patients with depression, a small but significant positive correlation was observed between SICI and ICF ratios (r_s = 0.21, p = 0.021), indicating a possible association between impaired cortical inhibition and elevated cortical excitation in depression. By contrast, the correlation between SICI and ICF was not observed in healthy adults (r_s = −0.08, p = 0.51).

Discussion

To our knowledge, this is the first TMS study investigating cortical inhibition and excitation in older adults with depression. Compared with younger healthy adults (56.1% mean cortical inhibition), we demonstrated that cortical inhibition is reduced in (i) older adults with early- or late-onset depression, (ii) older healthy adults and (iii) younger adults with depression (35.6, 39.7, and 41.7% mean inhibition, respectively). The diminished cortical inhibition observed in these three groups was specific to SICI, which reflects GABA_A receptor-mediated inhibition, and not to CSP, which reflects GABA_B receptor-mediated inhibition. The observed SICI reductions are congruent with earlier reports of lower SICI in adults with depression by Bajbouj et al. (left hemisphere: 35.2% inhibition, compared with 69.3% inhibition in healthy controls) [30] and Lefaucheur et al. (left hemisphere: 36% inhibition, compared with 77.4% in healthy controls) [31]. Contrary to our hypotheses, we did not discern any differences in cortical inhibition (SICI and CSP) between older adults with depression and age-matched healthy adults. However, our findings of (i) reduced GABA_A receptor-mediated neurotransmission in older vs. younger adults and (ii) an association between decreasing inhibition and advancing age, are consistent with previous studies that have shown lower SICI in older, compared to younger, healthy adults [46, 47], and a similar linear relationship between SICI and age [47]. Incongruent reports of elevated SICI in older adults might be accounted for by small sample sizes in earlier reports and methodological differences between studies, including different sample demographics, CS intensities, ISIs, and locations of EMG measurement [33, 48]. Taken together, our findings suggest that GABA_A receptor-mediated inhibitory neurotransmission is diminished during both aging and depression, and that the influence of aging and depressive states on TMS measures of cortical inhibition cannot be differentiated in LLD.

Various biological factors may have contributed to the observed reductions in cortical inhibition with both advancing age and depression. First, we observed a weak negative correlation between cortical inhibition (SICI) and cortical excitation (ICF) in depressed patients; it is possible that the observed inhibition deficits involve an abnormal inhibition/excitation balance in adults with depression across the lifespan. Accordingly, an abnormal GABA/glutamate signaling balance has been shown in a rodent model of depression, accompanied by a shift in the GABA/glutamate balance with antidepressant treatment [49]. Although speculative, it is possible that an abnormal balance of cortical inhibitory and excitatory neurotransmission could contribute to the observed imbalance between resting-state functional networks in depressed patients, which in turn could drive the imbalance between internal and external mental contents that is observed in depression [50]. We also found that the burden of comorbid physical illness was weakly associated with reduced GABA_A receptor-mediated inhibitory neurotransmission in older adults with depression. It is important to consider the effects that systemic health in LLD may have on cortical inhibition and brain functioning in general. Depression is associated with higher rates of age-related illness [43, 44], and there is some evidence that age-related pathologies, such as hypoxia, can impair GABAergic inhibitory neurotransmission [51]. Depression has similarly been associated with accelerated biological aging [52]; aging can affect other biological processes, including muscle mass, motor units, and nerve conduction velocity [53–55], which may influence TMS measures of GABAergic neurotransmission in LLD. Likewise, stressful and traumatic experiences, which typically accumulate with advancing age and are risk factors for the development of depression [56], have been shown to influence GABAergic neurotransmission [57]. As such, the functional consequences of biological shifts associated with aging and depression are likely driven by the interaction of many different mechanisms, rather than GABAergic inhibition alone. Furthermore, deficits in TMS measures of cortical inhibition have been observed in a wide range of psychiatric disorders [58]. Thus, rather than being specific to depressive states, the deficits in cortical inhibition observed here in depression may be more generally associated with psychopathology across the lifespan.

In contrast to SICI, the present study did not find significant effects of age or depression on CSP duration. Changes in CSP have been reported following administration of benzodiazepines, which are positive modulators of GABA_A receptors [26]. CSP may therefore reflect the tonic conductance of a subtype of GABA_A receptors, in addition to GABA_B receptor-mediated neurotransmission, whereas SICI more likely reflects the phasic, i.e., transient, activation of GABA_A receptors at the synapse [59]. Nevertheless, abnormally short CSPs have been reported previously in older healthy adults [60] and younger patients with depression [30–32]. However, the mean CSP duration reported here in 30 younger healthy adults (134.2 ± 36.9 ms) appears to be considerably shorter than the mean CSP duration observed by our group previously in a sample of 25 healthy controls of similar age (158.5 ± 31.7 ms [32]). The lack of significant CSP differences between groups in the present study may be related to the shorter mean CSP in the current sample of healthy controls as compared to what was previously demonstrated. The lower healthy control CSP durations reported in the current study as compared with earlier reports could reflect the high between-subject variability of the CSP measure [61]; as such, caution should be taken when interpreting between-subject differences in CSP.

We similarly observed no significant differences in cortical excitation measures (RMT and ICF) between any groups. Our results are congruent with the majority of TMS studies assessing indices of phasic glutamatergic functioning, which have found similar cortical excitation in MDD patients and controls [30, 32, 58]. Notably, RMT is primarily dependent on ion channel conductivity, voltage-gated Na+/K+ channels [21, 62], and is unaffected by GABA [21], glutamate [63, 64], and dopamine [65]. By contrast, ICF is closely related to NMDA receptor-mediated glutamatergic function [21, 26]. NMDA receptors are a target in the treatment of depression [66], however not all patients benefit from treatment with NMDA antagonists [67]. Moreover, although some MRS studies have found abnormally low cortical concentrations of glutamate in depression [68], MRS measures of glutamate-related metabolites reflect a mix of physiological and non-physiological components of the glutamatergic system, rather than NMDA receptor-mediated cortical neurotransmission per se; as such, ICF does not correlate with MRS measures of glutamate concentrations [69]. Going forward, an investigation of the ability for cortical excitability to change, by indexing neuroplasticity using the TMS paradigm paired associative stimulation, may elicit a clearer understanding of pathophysiological mechanisms related to cortical excitation in LLD.

This work also complements the TMS literature investigating the influence of aging on various aspects of neurophysiology. We observed significant reductions in GABA_A receptor-mediated neurotransmission in the motor cortex with advancing age, yet interestingly no significant differences between younger and older adults in other TMS measures of cortical inhibition and excitation (CSP, ICF, and RMT). Previous studies have similarly found no significant differences in cortical excitation with age [46–48], and although earlier TMS studies of cortical inhibition in older healthy adults have yielded mixed results [34], the present work involved one of the largest samples of older adults tested in a single study to date. There is evidence that other aspects of motor cortex neurophysiology are diminished in older compared with younger adults, including neuroplasticity [70, 71], interhemispheric inhibition [72, 73], and inhibitory and excitatory connectivity between ipsilateral motor regions [74]. The functional consequences of specific changes in the physiology of the motor cortex have yet to be fully characterized, yet may shed some light into age-related decline in motor control [75]. For example, contrary to the current findings, a variant of the ICF paradigm that may specifically reflect excitatory processes prior to and during particular types of grasp has been shown to be abnormal in old age and to correlate with hand dexterity [76]. Moreover, factors that can normalize impaired cortical physiology, such as an active lifestyle [73], may be useful in preventing age-related decline in functioning.

Of note, the majority of older adults with depression studied here were treatment-resistant (75%). Some studies have implicated greater GABAergic deficits in patients with treatment-resistant depression, as compared with non-treatment-resistant patients [32, 77]. However, our findings of reduced SICI in older adults with depression, as compared with younger healthy controls, were observed for subgroups of treatment-resistant and non-treatment-resistant older patients alike (see Supplementary Materials and Methods).

Some limitations of our study should be considered. First, the presented TMS findings from the motor cortex may not directly reflect brain regions involved in the pathology of depression. Future studies of LLD neuropathology should assess cortical inhibition directly in the DLPFC, using TMS combined with EEG [78]. Second, although TMS currently offers the only in vivo measures of phasic GABAergic activity in humans, TMS measures of cortical inhibition are indirect indices of GABAergic inhibitory neurotransmission. In addition, future work should include a more comprehensive clinical characterization of the control groups, such as inclusion of CIRS, MMSE, and depression scores, and further neuropsychological characterization of the LLD patients. At last, owing to our inclusion of older adults with depression that presented with cardiac or cerebrovascular risk factors, it is possible that some of the patients studied here suffered from vascular depression. Although several clinical studies show a strong correlation between LLD and cardiac risk factors, there is evidence that patients with vascular depression have different neurophysiological profiles than patients with nonvascular MDD [79]. However, the majority of the patients in the current study had recurrent MDD with onset before the age of 60, whereas vascular depression is more closely associated with late-onset MDD (onset ≥ 60 years) [80, 81].

Overall, our findings, taken together with previous reports of impaired cortical inhibition in younger adults with depression, are consistent with the age-by-disease model of LLD, which posits that age-related biological changes that contribute to the impairment of specific neuronal and glial processes promote vulnerability to the development of depressive symptoms in late-life [6, 7]. Previous literature reveals a robust and heterogeneous effect of age on numerous biological pathways that overlap with those implicated in depression pathology, including, but not limited to, glial-mediated inflammation, oxidative stress, and calcium regulation [6]. There is evidence that physiological changes in the brains of older adults and the brains of younger depressed adults are similar. Contributions of genetic, endocrine, epigenetic, and environmental factors may determine whether and when pathophysiological thresholds are reached and depressive symptoms are expressed [7].

Alternatively, the current finding that reductions in cortical inhibition are similar in depression and old age are in line with accumulating evidence that depression is a disease of accelerated biological aging [52]. Major depressive disorder is associated with higher rates of medical comorbidities, including heart disease [43], metabolic syndrome [82], and Alzheimer’s disease [83], which commonly occur in old age. Moreover, cellular biomarkers of aging and age-related disease have been observed in depression. For example, shorter telomere length, which promotes apoptosis and is typically observed in old age, has been associated with depression [84]. More specifically, an increase in the number of toxic processes, such as oxidative stress, coupled with a decline in protective factors, such as antioxidants, (i) can cause cellular damage, and (ii) have been implicated in both age-related diseases and depression [52]. Deficits in cortical inhibitory neurotransmission in depression, shown here to mimic reductions that occur in old age, therefore lend further support for the involvement of premature biological aging in depression.

In conclusion, our study is the first to examine TMS measures of cortical inhibition and excitation in LLD. We investigated a relatively large sample of older adults with early- or late-onset depression, and one of the largest samples of healthy older adults to date. Our findings indicate that cortical inhibition measures cannot differentiate between the changes associated with aging or with depression in late life. However, our findings of SICI reduction (i.e., cortical inhibition deficits), taken together with comparable previous findings of SICI deficits in younger patients with depression [30, 31], suggest that both advancing age and MDD involve reduced GABAergic inhibitory neurotransmission. These overlapping cortical inhibition reductions mirror other biological changes that occur with both age and depression, such as changes in the expression of genes coding for GABA interneuron-related peptides [85]. These observations are consistent with the conceptualization of depression as a disease of accelerated aging, and with the age-by-disease model of LLD [6, 7] and may provide targets for both the prevention and treatment of LLD. Future research should aim to clarify whether reductions in GABAergic cortical inhibition in old age, as observed here, constitute a biological risk factor for the development of depression in late-life.

Competing interests

JIL, AB, and RZ report no biomedical interests or conflicts. BHM currently receives research funding from Brain Canada, the Centre for Addiction and Mental Health (CAMH) Foundation, the Canadian Institutes of Health Research (CIHR), Patient-Centered Outcomes Research Institute (PCORI), and the US National Institute of Health (NIH). During the last five years, he also received research support from Bristol-Myers Squibb (medications for a NIH-funded clinical trial), Eli-Lilly (medications for a NIH-funded clinical trial), and Pfizer (medications for a NIH-funded clinical trial). He directly own stocks of General Electric (less than $5000). EJL reports research funding (current/past) from Janssen, Alkermes, Acadia, Takeda, Lundbeck, Barnes Jewish Foundation, PCORI, and Taylor Family Institute for Innovative Psychiatric Research. CFR has received research support from the NIH, PCORI, the Center for Medicare and Medicaid Services, the American Foundation for Suicide Prevention, the Brain and Behavior Research Foundation, and the Commonwealth of Pennsylvania. Bristol Meyers Squib and Pfizer have provided pharmaceutical supplies for his NIH sponsored research. JFK received medication supplies from Indivior to support this investigator-initiated trial. He has also received medication supplies from Pfizer for investigator-initiated work. JFK receives research funding from NIH and PCORI. TKR receives research support from Brain Canada, Brain and Behavior Research Foundation, Canada Foundation for Innovation, the CIHR, Ontario Ministry of Health and Long-Term Care, Ontario Ministry of Research and Innovation, the NIH, and the W. Garfield Weston Foundation. YN has received equipment-in-kind support for an investigator-initiated study from Magventure Inc. He has also received research support from Otsuka Pharmaceutical Co., Ltd, Shionogi & Co., Ltd., and Meiji Seika Pharma Co., Ltd. YN has received research support from Meiji Yasuda Mental Health Foundation. ES has received support from NIH, CIHR, the Brain & Behavior Research Foundation (NARSAD) and the Campbell Family Mental Health Research Institute of CAMH. ES is co-inventor on a US provisional patent application that covers compounds modulating the function of GABA neurons. ZJD has received within the last 3 years both research and equipment in-kind support for an investigator-initiated study through Brainsway Inc and Magventure Inc. ZJD has also received monies for participation on an advisory board from Sunovion Inc. Finally, ZJD owns > $10,000 (CAD) in stock of Biogen Inc. DMB has received research support from CIHR, NIH, Brain Canada and the Temerty Family through the CAMH Foundation and the Campbell Research Institute. He receives research support and in-kind equipment support for an investigator-initiated study from Brainsway Ltd. and he is the site principal investigator for three sponsor-initiated studies for Brainsway Ltd. He also receives in-kind equipment support from Magventure for an investigator-initiated study. He receives medication supplies for an investigator-initiated trial from Indivior.

Electronic supplementary material

Supplementary Information accompanies this paper at (10.1038/s41386-018-0093-x).