Treatment-Resistant Depression: The Cause-First Workup Before TMS

If you are in acute danger, contact emergency services immediately or call your national crisis line.

Roughly one in three patients with major depressive disorder does not reach remission after two adequate antidepressant trials. The STAR*D trial, which followed 4,041 real-world patients through up to four treatment steps, showed that remission rates fell sharply with each additional step — from 33 % on first-line citalopram to under 14 % by step four — and that relapse was the rule, not the exception, even among those who eventually responded (Rush et al., PMID 17074942). By the time a patient reaches a specialist with the label “treatment-resistant,” they have typically been depressed for years, have tried three to six medications, and have been told their brain is broken.

It isn’t. What is broken — almost always — is the diagnostic process that never looked beyond the drug-symptom matching model.

What “treatment-resistant” actually means

The standard TRD definition — failure on two adequate trials — is a treatment history, not a diagnosis. It tells you what hasn’t worked. It says nothing about why. That gap is where years of unnecessary suffering live.

Clinical experience at this practice, and a growing body of research, points consistently to four categories of missed causation:

1. Brain network dysfunction that cannot be identified without mapping
2. Systemic inflammation driving or perpetuating depressive neurochemistry
3. Occult infections generating chronic immune activation and neuropsychiatric symptoms
4. Relational toxicity — abuse, manipulation, or narcissistic control — that resets the nervous system into permanent threat mode

None of these appear on a standard DSM-5 checklist. All of them respond to targeted intervention. The workup below runs through each domain before any escalation decision is made.

Domain 1: Brain network mapping with qEEG

Standard EEG sees seizures. Quantitative EEG (qEEG) with LORETA source localization sees functional circuit architecture — specifically, which networks are hypoactive, hyperactive, or dysrhythmic in a living, awake brain. That matters because depression is not one circuit problem: anterior cingulate hypoactivity looks different from prefrontal hypoactivation, and each has a distinct response profile to rTMS frequency and target site.

Vlcek and colleagues (2020, PMC7057228) demonstrated that pre-treatment eLORETA source analysis separated rTMS responders from non-responders with reasonable accuracy in a small pilot cohort: responders showed significantly lower alpha-2 and beta-1 source current density in the left middle frontal gyrus at baseline. The finding reflects what we see clinically — the patients most likely to respond to left-DLPFC high-frequency rTMS are those whose qEEG shows measurable hypoactivation there to begin with.

What we look for in the qEEG:

• Left DLPFC alpha excess (linked to anhedonia-dominant depression, responds to 10 Hz rTMS)
• Anterior cingulate theta hyperactivation (OCD-spectrum depression, responds to 1 Hz right DLPFC or cingulate targeting)
• Diffuse delta slowing (may indicate subcortical pathology, infection, or metabolic disruption)
• Coherence abnormalities suggesting disrupted network connectivity

If qEEG reveals diffuse slowing or unexpected subcortical patterns, that is a flag to pursue Domains 3 and 4 before TMS rather than alongside it.

Domain 2: Inflammatory workup — CRP, cytokines, and the immune-depression axis

The inflammation–depression relationship is one of the most replicated findings in biological psychiatry. A 2024 transcriptomic analysis of the BIODEP cohort stratified MDD patients by CRP and by antidepressant response, finding that immunomodulatory and neuroprotective pathway expression distinguished responders from non-responders (PMID 39271754). A 2022 systematic review of 56 studies found that roughly one-third of depressed patients carry a low-grade inflammatory phenotype, and that elevated baseline CRP consistently predicts worse antidepressant response and greater symptom severity.

This is not a minor statistical association. In a patient with elevated CRP and persistent fatigue, cognitive slowing, and anhedonia that has failed multiple SSRIs, the probability that inflammation is driving the clinical picture is high enough to investigate before prescribing drug number four.

Minimum inflammatory panel:

• High-sensitivity CRP (hs-CRP) — target < 1 mg/L; hs-CRP > 3 suggests active inflammation
• IL-6, TNF-alpha (cytokine panel where available)
• Complete blood count with differential — eosinophilia as a mast cell or allergic signal
• Ferritin — elevated in inflammatory states; depleted in iron-deficiency that mimics depression
• Thyroid panel (TSH, free T3, free T4, TPO antibodies) — Hashimoto’s is frequently missed

If inflammation is confirmed, the source investigation runs in parallel with Domain 3.

Domain 3: Occult infection screening

This is where conventional psychiatry most reliably fails treatment-resistant patients. The infections below are not rare edge cases — they are systematically under-diagnosed in psychiatric populations because their testing is imperfect and their neuropsychiatric presentations are heterogeneous.

Lyme disease and tick-borne co-infections

A 2024 scoping review in Cureus (PMID 38752040) documented depressive symptoms, anxiety, aggression, and suicidal ideation as common features of persistent Lyme disease, noting that neuropsychiatric symptoms tend to persist even after standard antibiotic courses. The same review highlighted that Lyme disease can present with symptoms indistinguishable from primary psychiatric illness — a mimicry that creates a diagnostic trap when serology is the only test used.

Standard ELISA/Western blot serology misses a substantial proportion of clinically active cases, particularly in later-stage disease where immune exhaustion can suppress antibody production. At this practice, we use a panel that includes:

• Two-tier serology (ELISA + IgG/IgM Western blot)
• CD57 NK cell count (immunosuppression marker in persistent Lyme)
• PCR for Borrelia in symptomatic patients with high prior probability
• Co-infection serology: Bartonella (a frequent driver of neuropsychiatric symptoms including anxiety, rage, and depersonalization), Babesia, Ehrlichia/Anaplasma
• Mycoplasma pneumoniae antibodies

For detailed diagnostic criteria and the difference between German and US testing standards, see Lyme disease testing: what the standard panel misses.

Reactivated herpesviruses

A 2023 review in Frontiers in Psychiatry documented that HHV-6 DNA is detected more frequently in post-mortem cerebellar tissue from major depression and bipolar cases than in controls, and that EBV reactivation has been associated with depressive symptoms in multiple study populations. A 2025 meta-analysis of 19 studies (n ≈ 94,000) found that herpesvirus reactivation was associated with depression with an odds ratio of approximately 1.99 for EBV specifically. Bidirectionality likely plays a role — chronic stress reactivates latent herpesviruses, and the resulting immune activation perpetuates depression.

Viral reactivation panel:

• EBV VCA IgM/IgG, EA-D IgG (early antigen — the reactivation marker)
• HHV-6 IgG with avidity testing
• CMV IgM/IgG
• HSV-1/HSV-2 IgG (less commonly a driver but documented in some neuropsychiatric presentations)

Positive reactivation markers do not automatically mean antiviral treatment is indicated. They mean the investigation continues: quantify viral load where possible, assess NK cell function, and determine whether immune restoration (thymosin alpha-1, targeted peptide support) or direct antiviral therapy is the priority.

Domain 4: The relational environment

This domain is frequently omitted from workups because it feels outside the biomedical frame. It shouldn’t be. A nervous system that is chronically exposed to abuse, manipulation, humiliation, or narcissistic invalidation is a nervous system running a continuous threat response. No antidepressant can overcome that. No TMS protocol will produce lasting remission if the patient returns home to the same environment after every session.

The clinical questions here are direct:

• Does the patient describe a relationship (partner, parent, employer) where they consistently feel diminished, afraid, confused, or responsible for the other person’s emotional state?
• Are there cycles of idealization and devaluation, isolation from support networks, financial control?
• Does the patient’s affect or autonomic state visibly shift when the subject of that person is raised?

This is not soft medicine. Persistent interpersonal stress is a measurable biological event: elevated cortisol, chronically elevated CRP, impaired sleep architecture, HPA axis dysregulation. Identifying it and addressing it — whether through individual trauma therapy, couples work where appropriate, or supported separation — is a treatment decision with clinical relevance comparable to adjusting a medication dose.

Putting the domains together: the pre-TMS workup sequence

This is not a checklist to run in parallel while scheduling TMS. It is a sequential investigation where findings in one domain redirect the others:

1. History and qEEG — establish the circuit phenotype and identify red flags for systemic pathology
2. Inflammatory panel — if CRP > 3 or cytokines elevated, pursue infection workup urgently
3. Infection panel — Lyme/co-infections first (most commonly missed), then viral reactivation
4. Relational history — screen at every visit; this domain often opens only when the patient feels heard
5. TMS targeting — only after active infections are under treatment, inflammation is trending down, and relational toxicity is acknowledged and being addressed

The TMS that follows a completed workup is different in character from TMS applied as a symptom-suppression tool. It is working with a cleaned-up system rather than against an ongoing pathological tide.

A note on ketamine and esketamine

Ketamine and esketamine have genuine roles in this population: as time-limited bridges for patients in acute suicidal crisis while the workup proceeds, or as short-term adjuncts to TMS induction in patients with severe anhedonia whose dopaminergic circuits need a reset before they can respond to repetitive magnetic stimulation. Neither is a stand-alone strategy. The antidepressant effect of a ketamine infusion lasts days to weeks; using it as a maintenance treatment while fundamental causes remain unaddressed is using a pain medication to avoid diagnosing the fracture.

For the clinical protocol combining ketamine and rTMS, see Ketamine infusion for depression: what the evidence shows.

The ECT question

Patients who have been labeled treatment-resistant are frequently offered electroconvulsive therapy as the logical endpoint of the escalation ladder. Julian’s position is unambiguous: ECT should have no place in the treatment of depression and should be forbidden without a patient’s fully informed consent; every course should require a judge’s approval and an independent reviewer before proceeding.

This position rests on the evidence. Mathiassen and colleagues (2026, British Journal of Psychiatry, PMID 40357797) have demonstrated that ECT can permanently erase autobiographical memories — not a temporary side effect, a documented permanent loss. And the efficacy argument for ECT has been undercut: Terao et al. (2025, PCN Reports, PMID 40590032) conducted a network meta-analysis of 35 RCTs and found no significant difference in response rate, remission rate, or tolerability between IV ketamine, rTMS, and ECT for stage-2-or-higher TRD. The more tolerable alternatives work equally well.

There is no clinical scenario in which skipping the workup described above and proceeding directly to ECT is the right path. For a full review of the evidence and the case against ECT, see ECT side effects and what the alternatives actually show.

Who is this workup for?

The protocol above applies to any patient who meets the standard TRD definition (failure of two adequate antidepressant trials) and who has not had a systematic biological investigation. In practice, that is most of them.

For patients where suicidal ideation requires faster remission than a standard 4–6 week TMS course can deliver, accelerated TMS protocols — SWIFT and 5×5 — have 2026 evidence supporting their use.

The workup is especially urgent when the clinical picture includes:

• Fatigue disproportionate to mood severity
• Cognitive slowing, word-finding difficulty, or brain fog
• Joint pain, neurological symptoms, or a history of outdoor exposure in Lyme-endemic areas
• History of childhood or current interpersonal trauma
• Inflammatory markers above reference range on routine labs
• Previous partial responses that failed to consolidate

Related Articles

• ECT side effects and alternatives: what the evidence shows
• TMS for depression: what 20 years of evidence shows
• Lyme disease testing: what the standard panel misses
• CFS complete guide: cause-first approach to chronic fatigue syndrome
• Ketamine infusion for depression: protocol and evidence