Epilepsy Diagnostic Workup
Epilepsy Diagnostic Workup
What Do You Need to Know?
- EEG is the single most important ancillary test — routine EEG detects IEDs in ~25–35% on first study; cumulative yield approaches 80–90% in many series with 3–4 repeated studies including sleep (figures vary by population and whether sleep is included)
- MRI with epilepsy protocol at 3T is preferred (HARNESS-MRI 2019); incremental yield over a dedicated 1.5T epilepsy protocol is ~10–15%, larger when compared with non-dedicated 1.5T. A high-quality 1.5T epilepsy protocol is acceptable when 3T is unavailable (especially FCD, hippocampal sclerosis)
- First seizure labs: glucose, BMP (Na, Ca, Mg), CBC, tox screen — identify provoked causes before committing to ASM therapy
- Surgical evaluation after failure of 2 appropriate ASMs — video-EEG, FDG-PET, ictal SPECT, MEG, neuropsychological testing
- Autoimmune workup when new-onset refractory seizures, FBDS, limbic encephalitis features, or APE2 ≥4 — send serum AND CSF panels
- Genetic testing highest yield in neonatal/infantile seizures and DEEs (Dravet >80%, BFNS ~80%, DEE 30–50%)
- Prolactin elevated post-GTCS and focal impaired awareness (not absence, not PNES) — limited sensitivity; draw 10–20 min AFTER the event and compare to baseline drawn ≥6 h later
🚩 Don’t Miss — Test-Day Priorities
- Routine 30-min EEG sensitivity ~30–50% in known epilepsy: a single normal EEG does NOT exclude epilepsy — epilepsy remains a clinical diagnosis
- Sleep-deprived EEG increases yield to ~80%: sleep deprivation activates IGE and focal IEDs — order this when routine EEG is nondiagnostic but suspicion remains high
- MRI epilepsy protocol = 3T with thin-slice coronal hippocampi + FLAIR + T1 IR + GRE/SWI: looks for hippocampal sclerosis, FCD, cavernoma, DNET/ganglioglioma, encephalomalacia — standard brain MRI is insufficient
- Hippocampal sclerosis triad: volume loss + T2/FLAIR hyperintensity + loss of internal architecture on coronal T2 perpendicular to hippocampus
- FDG-PET shows interictal HYPOmetabolism at the epileptogenic zone (NOT hypermetabolism) — classic board trick
- Ictal SPECT: Tc-99m HMPAO injected during seizure (ideally <30 sec from onset) vs baseline interictal; co-registered with MRI = SISCOM for localization
- Video-EEG LTM is essential for PNES vs epileptic: capture of the habitual spell with no EEG correlate AND incongruent semiology — confront with multidisciplinary team
- cEEG for ICU/altered MS to detect NCSE — subclinical seizures in 10–30% of acute brain injury patients
- Genetic testing for DEEs: gene panel first (Dravet SCN1A, CDKL5, STXBP1, KCNQ2, KCNT1, PCDH19); WES if panel negative; CSF glucose for GLUT1 (SLC2A1)
- Do NOT misread benign variants as epileptic: wicket spikes, BETS/SSS, 14-and-6 positive spikes, 6-Hz phantom spike-wave, SREDA, BETS-of-sleep — all are normal variants
🔍 Buzzwords & Pathognomonic FindingsEEG patterns · MRI / imaging · Workup pearls
EEG patterns / findings
- Spike, sharp wave, spike-wave, polyspike-wave → epileptiform discharges (define lateralization/localization)
- 3 Hz generalized spike-wave activated by HV → childhood absence epilepsy (CAE)
- Photoparoxysmal response (PPR) — generalized SW to photic stim → IGE / photosensitive epilepsy (esp JME); 5% normal but pathologic if seizure occurs
- Hypsarrhythmia → infantile spasms (West syndrome)
- Centrotemporal sharps activated by sleep → self-limited epilepsy with centrotemporal spikes (BECTS)
- Wicket spikes / BETS-SSS / 14-and-6 / 6-Hz phantom SW / SREDA → BENIGN variants — do NOT call epileptic
- Captured habitual spell with NO EEG correlate → PNES (especially if incongruent semiology)
- NREM sleep recording → maximally activates IEDs (REM suppresses)
MRI / imaging
- Coronal T2/FLAIR hippocampal atrophy + signal hyperintensity + loss of internal architecture → hippocampal sclerosis (mTLE)
- Transmantle sign on FLAIR (linear hyperintensity cortex-to-ventricle) → focal cortical dysplasia type IIb (Taylor type, balloon cells)
- Popcorn lesion with hemosiderin rim on SWI/GRE → cavernoma (cavernous malformation)
- Cortical/subcortical “bubbly” T2-hyperintense temporal lobe mass in young patient → DNET
- Cystic temporal lobe mass with enhancing mural nodule → ganglioglioma
- Interictal FDG-PET HYPOmetabolism at seizure focus → epileptogenic zone (80–90% mTLE; 45–60% extratemporal)
- Ictal SPECT hyperperfusion subtracted from interictal & overlaid on MRI → SISCOM localization
- Encephalomalacia / gliosis on FLAIR → post-traumatic / post-stroke epilepsy
Workup / pitfalls
- Single normal routine EEG → does NOT exclude epilepsy — repeat with sleep / sleep deprivation
- SEEG (depth electrodes) → deep/bilateral/multilobar hypotheses; lower morbidity than grids; no large-area cortical mapping
- Subdural grids → superficial neocortical focus needing extensive language/functional mapping
- MEG dipole localization → superior for deep/tangential sources; cortical mapping
- CSF/serum glucose ratio <0.4 (fasting LP) → GLUT1 deficiency (SLC2A1) — treat with ketogenic diet
- IV pyridoxine 100 mg with dramatic seizure cessation on EEG → pyridoxine-dependent epilepsy (ALDH7A1); trial PLP for PNPO
- HV-induced posterior slowing in adolescent → NORMAL response — not absence
First Seizure Evaluation: Children vs. Adults
| Component | Children | Adults |
|---|---|---|
| History | Witness/video; birth hx (HIE, prematurity); developmental milestones; febrile seizure hx; family hx; vaccination timing | Witness/video; prior unrecognized events (staring, morning myoclonus); alcohol/drug use; sleep deprivation; driving/occupational risk |
| Labs | Glucose, BMP (Na, Ca, Mg), CBC; metabolic workup in neonates (lactate, ammonia, amino acids); tox screen in adolescents | Glucose, BMP (Na, Ca, Mg), CBC, LFTs, ammonia, tox screen; prolactin (limited — draw 10–20 min after event, compare to baseline ≥6 h later) |
| When to CT | Emergent: trauma, focal deficits, persistent AMS, VP shunt; NOT routine for simple febrile seizure | Emergent: focal deficits, persistent AMS, trauma, anticoagulation, cancer hx, immunocompromised, meningeal signs |
| When to MRI | All focal seizures; DD + seizures; abnormal exam; NOT required if classic IGE with typical EEG (e.g., CAE with 3 Hz spike-wave) | ALL adults with unprovoked seizures — epilepsy protocol; 3T preferred (HARNESS-MRI 2019), dedicated 1.5T epilepsy protocol acceptable when 3T unavailable; CT alone is insufficient |
| EEG timing | Within 24–48 h; first routine EEG with sleep captures IEDs in ~50–60% of children; yield drops substantially without sleep — sleep recording essential in children | Within 24–48 h; IED ~25–35% on first routine EEG; sleep-deprived or prolonged EEG if initial nondiagnostic |
| EEG yield | Higher yield than adults; pathognomonic patterns (hypsarrhythmia, 3 Hz spike-wave, centrotemporal spikes) | Yield 80–90% after 3–4 routine EEGs with sleep deprivation and longer recording |
💎 Board Pearl
- EEG within 24–48 h has the HIGHEST yield — IEDs are most likely captured in the early postictal period
- Classic IGE in children (CAE, JAE, JME) with typical EEG may NOT require MRI — any atypical feature mandates imaging
When to Order EEG
Routine EEG (20–40 min)
- First-line test for any suspected seizure or epilepsy
- IED detection: ~25–35% on first routine EEG; cumulative yield approaches 80–90% in many series with 3–4 repeated studies including sleep (numbers vary by population and inclusion of sleep recordings)
- Should include wakefulness + drowsiness/sleep for maximal yield
- Normal EEG does NOT exclude epilepsy — epilepsy is a clinical diagnosis
Sleep-Deprived EEG
- Sleep-deprived EEG adds ~20% incremental yield over routine awake EEG
- Recording during NREM sleep (regardless of sleep deprivation) activates most IEDs — capturing sleep is the key driver of yield
- Particularly useful for: suspected IGE (JME, CAE), temporal lobe epilepsy, nondiagnostic routine EEG
Prolonged / Ambulatory EEG (24–72 h)
- When routine EEG is nondiagnostic but clinical suspicion remains high
- Captures interictal and potentially ictal events; correlates symptoms with EEG changes
- Home ambulatory EEG is an alternative to inpatient monitoring for selected patients
Continuous EEG (cEEG) — ICU Indications
- Unexplained altered mental status / encephalopathy
- Post-cardiac arrest (detect nonconvulsive SE; prognostication)
- Refractory SE (titrate IV anesthetics to electrographic seizure suppression; burst suppression is commonly used in deeper coma but is not the only evidence-based endpoint)
- Comatose patients with suspected subclinical seizures
- Acute brain injury (TBI, SAH, ICH) — subclinical seizures in 10–30%
- Per ACNS 2015: ≥24 h cEEG in non-comatose patients with unexplained altered mental status; ≥48 h in comatose patients (yield rises from ~50% at 24 h to ~80–90% at 48 h in coma)
Activation Procedures
| Procedure | Mechanism | Key Clinical Utility |
|---|---|---|
| Hyperventilation (3–5 min) | Hypocapnia → vasoconstriction → hyperexcitability | Provokes absence seizures >90%; activates 3 Hz spike-wave in CAE |
| Photic stimulation | Intermittent flashing light (1–30 Hz) | PPR: generalized spike-wave; seen in IGE (especially JME). Distinguish from photic driving |
| Sleep | NREM disinhibits cortical networks | NREM activates most IEDs; REM suppresses them. Adds 20–30% yield |
Clinical Pearl
Hyperventilation is the single best activation procedure for absence epilepsy — triggers 3 Hz spike-wave in >90% of untreated CAE. If a board question describes staring spells with a normal interictal EEG, the answer is hyperventilation during the EEG, not a repeat study.
MRI Epilepsy Protocol
ILAE HARNESS-MRI Protocol (Bernasconi et al. 2019)
- 3T preferred — incremental yield over a dedicated 1.5T epilepsy protocol is ~10–15%, larger when compared with non-dedicated 1.5T. A high-quality 1.5T epilepsy protocol is acceptable when 3T is unavailable
- Three core sequences: 3D T1 (1 mm isotropic), 3D FLAIR (1 mm isotropic), high-resolution 2D coronal T2 perpendicular to hippocampus
| Sequence | Key Findings | Primary Targets |
|---|---|---|
| 3D T1 MPRAGE | Volumetric analysis; cortical thickness mapping | FCD (blurred gray-white junction); hippocampal volumetry; VBM post-processing |
| 3D FLAIR | Hyperintense signal in abnormal cortex/hippocampus | HS (FLAIR hyperintensity + atrophy); FCD (transmantle sign); tumors; gliosis |
| Coronal T2 (perp. to hippocampus) | Hippocampal architecture; size/signal asymmetry | HS detection — atrophy + T2 hyperintensity; most sensitive plane |
| SWI / GRE | Blooming from hemosiderin, calcium | Cavernomas; calcified lesions; Sturge-Weber; prior hemorrhage |
| Post-contrast T1 (if indicated) | Enhancement pattern | Tumors (DNET, ganglioglioma); infection; leptomeningeal disease |
💎 Board Pearl
- 3T is preferred (HARNESS-MRI 2019) — incremental yield over a dedicated 1.5T epilepsy protocol is ~10–15%, larger when compared with non-dedicated 1.5T. A high-quality 1.5T epilepsy protocol is acceptable when 3T is unavailable. FCD and HS most commonly missed at lower field strength
- Transmantle sign on FLAIR (linear hyperintensity cortex-to-ventricle) = pathognomonic for FCD type IIb (balloon cells)
- MRI-negative epilepsy = 20–30% of drug-resistant cases — MAP (morphometric analysis program), VBM, or 7T can reveal additional lesions in MRI-negative cases
Specialty Workup Decision Table
| Indication | Workup | Key Tests | When to Order |
|---|---|---|---|
| Surgical Evaluation | Video-EEG | Ictal onset localization; semiology; confirm epileptic vs. nonepileptic | After failure of 2 ASMs (drug-resistant epilepsy); refer early — average delay 15–20 years |
| 3T MRI | Structural lesion (HS, FCD, tumor, vascular malformation) | ||
| FDG-PET | Interictal hypometabolism; 80–90% for mTLE; 45–60% extratemporal | ||
| Ictal SPECT (SISCOM) | Ictal hyperperfusion; inject tracer as early as possible, ideally <30 sec, no later than 60 sec from clinical/electrographic onset | ||
| MEG | IED dipole localization; superior for deep/tangential sources | ||
| Neuropsych testing | Cognitive baseline; lateralization; predicts postop outcome | ||
| fMRI / Wada | fMRI replaces Wada for language lateralization (>90% concordance in typical right-handers); Wada still indicated for memory lateralization before dominant mesial temporal resection, and when fMRI language inconclusive/atypical | ||
| Intracranial EEG | SEEG (deep, bilateral, or multilobar hypotheses; lower hemorrhage/infection risk; no contiguous cortical mapping) vs subdural grids (superficial neocortical focus requiring large-area functional/language mapping; higher morbidity) | ||
| Autoimmune Workup | Serum antibody panel | NMDAR, LGI1, CASPR2, GABA-B, AMPAR, DPPX, GAD65 | New-onset refractory seizures; FBDS (→ LGI1); limbic encephalitis; APE2 ≥4; subacute cognitive decline |
| CSF antibody panel | NMDAR (CSF > serum sensitivity); OCBs; cytology | ||
| Body CT or PET/CT | Teratoma (NMDAR), thymoma (CASPR2/LGI1), SCLC (GABA-B) | ||
| MRI brain | Mesial temporal FLAIR signal (limbic encephalitis); may be normal early | ||
| Genetic Testing | Epilepsy gene panel | 100–500+ genes; fastest/cheapest first-line genetic test | Neonatal/infantile seizures; DEE; family hx; dysmorphic features; seizures + ID; MRI-negative DRE in young |
| WES (whole exome) | Broader coverage; trio analysis; yield 30–50% in DEE | ||
| WGS (whole genome) | Noncoding variants, structural variants, repeat expansions | ||
| Yield by syndrome | Dravet: >80%; BFNS: ~80%; DEE: 30–50%; focal: 10–20%; IGE: 5–10% | ||
| DD + Seizures | Chromosomal microarray | First-line for ID/DD + epilepsy; CNVs; yield ~15–20% | Seizures + DD/ID; dysmorphic features; autism + epilepsy |
| Fragile X | FMR1 CGG expansion; most common inherited cause of ID in males | ||
| Metabolic screen | Urine organic acids, amino acids, acylcarnitine, lactate | ||
| Metabolic Workup | Lactate / pyruvate | L:P ratio >25 = mitochondrial; MELAS, PDH deficiency | Neonatal/infantile refractory seizures; seizures + FTT; metabolic acidosis; regression |
| Amino acids / organic acids | NKH, MSUD, propionic/methylmalonic aciduria | ||
| Acylcarnitine profile | Fatty acid oxidation defects; carnitine deficiency | ||
| CSF glucose + lactate | GLUT1: CSF/serum glucose <0.4; responds to ketogenic diet | ||
| Pyridoxine trial | IV pyridoxine 100 mg with EEG; dramatic cessation = ALDH7A1; also trial PLP for PNPO | ||
| Infectious Workup | CSF basic studies | Cell count, protein, glucose, Gram stain, culture | Fever + seizure; immunocompromised; persistent AMS; meningeal signs; suspected encephalitis |
| CSF PCR panel | HSV PCR (start acyclovir empirically); VZV, enterovirus, HHV-6, arboviruses | ||
| Additional CSF | AFB (TB); cryptococcal Ag; cytology; VDRL (neurosyphilis) |
Clinical Pearl
FDG-PET in surgical evaluation: Interictal PET shows hypometabolism at the seizure focus (reduced glucose uptake between seizures). For mTLE, sensitivity is 80–90% — the most useful adjunctive imaging when MRI is negative or equivocal. Extratemporal sensitivity drops to 45–60%. Remember: PET shows HYPOmetabolism interictally, not hypermetabolism.
💎 Board Pearl
- GLUT1 deficiency: CSF/serum glucose ratio <0.4 — LP must be fasting; treat with ketogenic diet (bypasses glucose transporter)
- Pyridoxine-dependent epilepsy (ALDH7A1): Give IV pyridoxine 100 mg to ANY neonate with refractory seizures — dramatic cessation is diagnostic + therapeutic
- NMDAR antibodies are more sensitive in CSF than serum — always send BOTH panels
First-Line Labs for Acute Seizure — Quick Reference
| Test | Purpose | Key Thresholds / Notes |
|---|---|---|
| Glucose | Hypo/hyperglycemia | <40 mg/dL = provoked seizure; nonketotic hyperglycemia → focal seizures |
| Sodium | Hyponatremia | <120 mEq/L = seizure threshold; correct slowly (osmotic demyelination risk) |
| Calcium | Hypocalcemia | <7 mg/dL (total) or low ionized Ca; tetany, QTc prolongation |
| Magnesium | Hypomagnesemia | Lowers seizure threshold; coexists with hypocalcemia; treat Mg first |
| CBC | Infection screening; ASM baseline | Leukocytosis = infection or postictal stress response |
| LFTs | Hepatic encephalopathy; ASM baseline | Baseline before VPA, CBZ, PHT |
| Ammonia | Hepatic encephalopathy; urea cycle; VPA toxicity | Check in encephalopathic patients and those on valproate |
| Tox screen | Drug intoxication / withdrawal | Cocaine, amphetamines, synthetic cannabinoids, PCP; alcohol level |
| BUN/Cr | Uremia | Uremic encephalopathy causes seizures; affects ASM dosing |
Lumbar Puncture Indications
- Fever + seizure (especially new-onset in adults) — rule out meningitis/encephalitis
- Immunocompromised patient — broader infectious differential
- Persistent altered mental status not improving within 30–60 min postictal
- Meningeal signs (nuchal rigidity, Kernig/Brudzinski)
- Suspected autoimmune encephalitis (subacute onset, psychiatric features, movement disorder)
- CT before LP if signs of raised ICP (papilledema, focal deficits, obtundation)
Serum Prolactin
- Elevated (≥2x baseline) after: GTCS (most reliable) and focal impaired awareness
- NOT elevated after: absence seizures, simple partial seizures, PNES
- Draw 10–20 min AFTER the event; peaks at 15–20 min, returns to baseline by 1–2 h; compare to a baseline prolactin drawn ≥6 h later (or at the same circadian time) — a single post-ictal value is uninterpretable without baseline
- Sensitivity ~60% for GTCS — normal prolactin does NOT exclude epileptic seizure
- Most useful to distinguish GTCS from PNES; syncope can also transiently elevate prolactin
💎 Board Pearl
- Prolactin is a board favorite: elevated post-GTCS and focal impaired awareness, NOT in absence or PNES; draw 10–20 min AFTER the event, compare to baseline drawn ≥6 h later
- The ONLY lab that may help distinguish seizure types acutely — but do NOT rely on it to rule out seizures
Board Pearls — Epilepsy Diagnostic Workup
💎 Board Pearl
- EEG within 24–48 hours has highest yield — per ILAE 2014, epilepsy can be diagnosed after a single unprovoked seizure if recurrence risk is ≥60% over the next 10 years — most commonly satisfied by epileptiform IEDs on EEG, an epileptogenic lesion on MRI, or remote symptomatic etiology
- Hyperventilation provokes absence seizures >90% of the time — the single best activation procedure for CAE
- 3T is preferred (HARNESS-MRI 2019) — incremental yield over a dedicated 1.5T epilepsy protocol is ~10–15%, larger when compared with non-dedicated 1.5T. A high-quality 1.5T epilepsy protocol is acceptable when 3T is unavailable; FCD and HS most commonly missed at lower field strength
- FDG-PET = interictal hypometabolism (not hypermetabolism) at the seizure focus; 80–90% sensitivity for mTLE, 45–60% extratemporal
- GLUT1 = low CSF glucose, normal serum glucose (ratio <0.4) — treat with ketogenic diet
- Drug-resistant epilepsy = failure of 2 ASMs — refer for surgical evaluation; average delay is 15–20 years (far too long)
- APE2 score ≥4 triggers autoimmune testing — components: new-onset seizures, drug resistance, autonomic features, viral prodrome, inflammatory CSF, autoimmune MRI, psychiatric/cognitive features, malignancy hx
Clinical Pearls
Clinical Pearl
Ictal SPECT timing is critical: Inject radiotracer (Tc-99m HMPAO or ECD) as early as possible, ideally <30 sec and no later than 60 sec from clinical/electrographic seizure onset. Late injection captures postictal hypoperfusion instead of ictal hyperperfusion, leading to mislocalization. SISCOM (subtraction ictal SPECT co-registered to MRI) improves accuracy by subtracting interictal from ictal SPECT and overlaying the difference on MRI.
Clinical Pearl
Wada vs. fMRI: fMRI replaces Wada for language lateralization (>90% concordance in typical right-handers). Wada is still indicated for memory lateralization before dominant mesial temporal resection, and when fMRI language is inconclusive or atypical (e.g., left-handed, crossed dominance, prior early left-hemisphere injury).
Clinical Pearl
APE2 score for autoimmune epilepsy: A score ≥4 has reasonable sensitivity and should trigger antibody testing. Components include new-onset seizures, drug resistance, autonomic dysfunction, viral prodrome, inflammatory CSF, MRI suggesting autoimmune etiology, psychiatric/cognitive features, and malignancy history. This prevents unnecessary antibody testing in all epilepsy patients while capturing those most likely to have an autoimmune etiology.
References
- Krumholz A, Wiebe S, Gronseth GS, et al. Evidence-based guideline: management of an unprovoked first seizure in adults. Neurology. 2015;84(16):1705–1713.
- Fisher RS, Acevedo C, Arzimanoglou A, et al. ILAE official report: a practical clinical definition of epilepsy. Epilepsia. 2014;55(4):475–482.
- Bernasconi A, Cendes F, Theodore WH, et al. Recommendations for the use of structural MRI in epilepsy: ILAE Neuroimaging Task Force consensus. Epilepsia. 2019;60(6):1054–1068.
- Rosenow F, Lueders H. Presurgical evaluation of epilepsy. Brain. 2001;124(Pt 9):1683–1700.
- Graus F, Titulaer MJ, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol. 2016;15(4):391–404.
- Helbig I, Riggs ER, Barry CA, et al. ClinGen Epilepsy Gene Curation Expert Panel. Hum Mutat. 2018;39(11):1476–1484.
- Kwan P, Arzimanoglou A, Berg AT, et al. Definition of drug resistant epilepsy: ILAE consensus proposal. Epilepsia. 2010;51(6):1069–1077.
- Skidmore CT. Neuroimaging in epilepsy. Continuum (Minneap Minn). 2025;31(1, Epilepsy):61–80.
- Dubey D, Pittock SJ, Kelly CR, et al. Autoimmune encephalitis epidemiology and comparison to infectious encephalitis. Ann Neurol. 2018;83(1):166–177.
- Chen Z, Brodie MJ, Liew D, Kwan P. Treatment outcomes in newly diagnosed epilepsy: a 30-year longitudinal cohort study. JAMA Neurol. 2018;75(3):279–286.
Continue reading — sign in
The full note has more clinical pearls, tables, and board-focused tips. Free account, no fee.