Every IEM is, at root, a blocked or leaky step in a metabolic pathway. What determines the clinical face of that block — and whether it is treatable — is not the specific enzyme but what happens upstream and downstream of the lesion. Two questions organize the entire field: does a toxic intermediate pile up behind the block (intoxication), or does the pathway fail to deliver a needed product such as ATP (energy deficiency)? And is the offending molecule small and water-soluble (diffusible, dialyzable, often correctable) or large and structural (sequestered in an organelle, accumulating slowly and irreversibly)?

Small-molecule IEMs disturb the traffic of intermediary metabolites — amino acids, organic acids, ammonia, fatty acids, sugars. Because these molecules diffuse freely, equilibrate across compartments, and can be cleared by the kidney or by dialysis, small-molecule disease is the treatable end of the spectrum. It splits into two mechanistic camps:

• Intoxication — the substrate immediately proximal to the block accumulates to neurotoxic levels (leucine in MSUD, ammonia in urea cycle disorders, propionate/methylmalonate in the organic acidemias). The brain is normal at birth because the placenta clears maternal toxins; symptoms erupt only when the neonate begins to feed and catabolize, hence the classic symptom-free interval followed by acute encephalopathy.
• Energy deficiency — the pathway that should generate ATP or shuttle fuel fails (fatty-acid oxidation defects, the mitochondrial respiratory chain, GLUT1's import of glucose into brain). Here the trigger is the opposite metabolic state: fasting, when the body needs the broken pathway most.

Large-molecule IEMs are organelle-based storage disorders — lysosomal (MPS, gangliosidoses, Gaucher, Krabbe, MLD), peroxisomal (X-ALD, Zellweger), and the congenital disorders of glycosylation. A complex macromolecule cannot be degraded or assembled, so it accumulates within the organelle and distends the cell. There is no dialyzable toxin and no acute crisis to reverse; the damage is structural and cumulative, which is why these present as relentless, often irreversible neurodegeneration rather than episodic decompensation.

Why the tempo differs: This framework predicts the bedside picture. Intoxication and energy disorders cause episodic, catabolism- or fasting-provoked crises on a background that can look deceptively normal between events — and that reversibility is exactly what makes rescue possible. Storage disorders cause insidious regression because you cannot un-store a lysosome full of substrate. Roughly 80 metabolic disorders causing intellectual disability are actually treatable when caught early, which is the entire rationale for aggressive, time-sensitive workup (van Karnebeek & Stockler. 2012).

One gene, a spectrum of disease: Residual enzyme activity, not merely the presence of a mutation, sets severity. A near-null allele gives classic neonatal PKU or infantile Pompe; a leaky allele gives mild hyperphenylalaninemia or adult-onset limb-girdle Pompe. The same locus thus spans the catastrophic and the nearly silent — a reminder that genotype must always be read together with the biochemical phenotype.

Newborn screening exists because of the symptom-free interval. The intoxication and energy-deficiency IEMs do their damage in the first days to weeks of life, before a clinician would otherwise suspect anything — so a population test that flags the disorder during that silent window can convert a lethal or disabling condition into a manageable one. The logic is classic Wilson–Jungner public-health screening: the disorder must be serious, have a detectable preclinical phase, and have a treatment that works better when started early. Almost every condition on the panel meets that last criterion, which is what justifies screening an entire population to find a handful of affected infants.

Why tandem mass spectrometry changed everything: Older NBS used one assay per disorder (the Guthrie bacterial-inhibition test for PKU, for instance). MS/MS measures dozens of analytes — the amino acids and the acylcarnitines — from a single punch of a dried blood spot in a couple of minutes. A single run therefore captures the aminoacidopathies (elevated phenylalanine, leucine), the organic acidemias (abnormal short-chain acylcarnitines such as C3 in MMA/PA), and the fatty-acid oxidation defects (C8 in MCAD, C14:1 in VLCAD). One technology, many diseases — that economy is what made expanded screening feasible.

In the US the Recommended Uniform Screening Panel (RUSP) defines a national core set (>35 core plus several dozen secondary conditions); individual states add others, so the exact panel an infant receives is jurisdiction-dependent.

The conceptual limits clinicians must hold onto:

• A positive screen is a probability, not a diagnosis. MS/MS is tuned for high sensitivity, so it over-calls; confirmatory plasma amino acids, urine organic acids, enzymology, or molecular testing must follow before any dietary or pharmacologic intervention. False positives cluster in premature, TPN-fed, or transfused infants and with poor spot quality.
• A negative screen does not exclude an IEM. Disorders absent from the panel (most urea cycle disorders, many mitochondrial and storage diseases), milder alleles below the cutoff, and samples drawn too early can all screen normal. The screen tests for a defined list of metabolites — it is not a metabolic clearance certificate. A sick infant with a suggestive picture still warrants a full diagnostic workup regardless of the NBS result.

A small set of neurometabolic disorders carries a moral weight out of proportion to their frequency: the treatment is specific, often cheap, and the difference between giving it early and giving it late is the difference between a normal child and permanent disability or death. These are the don't-miss IEMs, and the reason to memorize them is mechanistic, not just clinical etiquette — each one is a place where a single intervention restores the missing function before the brain is irreversibly injured.

Why early matters so much here: In the intoxication disorders (MSUD, urea cycle, organic acidemias) the toxin is reversible while the patient is alive, so detoxification — protein cessation, IV glucose to halt catabolism, nitrogen scavengers, dialysis — can pull a child back from coma with the brain intact, but only if started before sustained edema and herniation. In the cofactor- and fuel-replacement disorders the logic is even cleaner: you are simply supplying what the lesion cannot make or transport.

• Pyridoxine-dependent epilepsy (ALDH7A1/antiquitin): the block in lysine degradation lets an aldehyde intermediate trap and inactivate pyridoxal-5'-phosphate, the active form of B6 and the cofactor for dozens of CNS enzymes including GAD. Replacing B6 restores the cofactor and stops otherwise-intractable neonatal seizures — which is why a pyridoxine trial belongs in every refractory neonatal seizure protocol.
• Biotinidase deficiency: the enzyme that recycles biotin fails, starving four carboxylases; oral biotin trivially refills the pool and prevents seizures, deafness, optic atrophy, and rash.
• GLUT1 deficiency (SLC2A1): the transporter that ferries glucose across the blood–brain barrier is haploinsufficient, so the brain is chronically fuel-starved despite normal blood glucose — the founding observation being persistent hypoglycorrhachia (low CSF glucose) with seizures and developmental delay (De Vivo et al. 1991). The CSF:blood glucose ratio runs <0.4 (normal >0.6). The fix bypasses the broken door entirely: a ketogenic diet supplies ketone bodies, which cross the barrier on a different transporter (MCT1) and feed neurons directly.
• Creatine deficiency (GAMT/AGAT/SLC6A8): the brain's phosphocreatine energy buffer is absent — recognizable as a missing creatine peak on MRS — and synthesis defects (GAMT, AGAT) respond to oral creatine.

The through-line is that for these conditions the diagnostic test is targeted and the therapy is mechanism-restoring, so the neurologist's job is simply to think of them — in any child with unexplained encephalopathy, refractory seizures, regression, or a movement disorder — and order the one assay that confirms it.

The diagnostic approach is driven by one biochemical fact: in the small-molecule disorders, the diagnostic metabolites are most abnormal during decompensation and may normalize between crises. Catabolic stress floods the blocked pathway, so an organic acidemia, urea cycle defect, or fatty-acid oxidation disorder can throw a near-normal screen once the child has recovered and is anabolic again. The single most important practical rule, therefore, is to draw the critical samples while the patient is sick — the acute crisis is a diagnostic opportunity, not just an emergency.

The first-tier panel maps onto the mechanistic categories. Each test interrogates a different compartment of intermediary metabolism, so the pattern across them localizes the lesion:

• Ammonia, blood gas, glucose, lactate are the rapid triage axis. Hyperammonemia without acidosis points to a urea cycle disorder; hyperammonemia with high anion-gap acidosis points to an organic acidemia (where accumulating organic acids secondarily inhibit ureagenesis); hypoglycemia with inappropriately low ketones suggests a fatty-acid oxidation defect; a high lactate with an elevated lactate:pyruvate ratio points to the respiratory chain.
• Plasma amino acids read the aminoacidopathies and urea cycle directly (phenylalanine in PKU, leucine in MSUD, citrulline/argininosuccinate placing the urea-cycle block, glycine in NKH).
• Urine organic acids (GC-MS) capture the organic acidemias and many energy disorders — methylmalonate, propionate, glutarate, and the orotic acid that distinguishes OTC deficiency from more proximal urea-cycle blocks.
• Plasma acylcarnitine profile fingerprints fatty-acid oxidation by chain length (C8 in MCAD, C14:1 in VLCAD, C16-OH in LCHAD).
• CSF studies are indispensable when the lesion lives behind the blood–brain barrier and the blood is normal: CSF glucose for GLUT1, CSF:plasma glycine ratio for non-ketotic hyperglycinemia, and CSF neurotransmitter metabolites/pterins for the dopamine–serotonin synthesis defects. These are the disorders a serum-only workup will miss.

Two cognitive traps deserve emphasis. First, a normal MRI does not exclude an IEM — GLUT1, creatine deficiency, NKH, and early intoxication disorders can all have unremarkable early imaging. Second, exhaustive sequential single-gene or single-enzyme testing is obsolete: once the first-tier biochemistry is collected (ideally during crisis) and a treatable signal is excluded or addressed, an unrevealing workup should proceed promptly to whole-exome or whole-genome sequencing rather than years of one-at-a-time assays.

Every IEM therapy is an answer to the same question the disease poses — too much of a toxin, or too little of a product? — and the menu of strategies maps directly onto the mechanistic categories rather than onto individual diseases.

Reduce the load behind the block. When the problem is accumulation, you cut what flows into the pathway. Dietary restriction removes the precursor (phenylalanine in PKU, branched-chain amino acids in MSUD, lysine in GA1, galactose in galactosemia), and substrate reduction therapy uses a small molecule to throttle synthesis upstream (miglustat/eliglustat dialing down glucosylceramide production in Gaucher and NPC). In the urea cycle, nitrogen scavengers open an alternative exit route for the toxin (ammonia) rather than fixing the broken enzyme.

Replace or boost what is missing. When the lesion is a deficient product or a sluggish enzyme, you supply it. Cofactor supplementation is the highest-yield, lowest-cost intervention in all of metabolic medicine — pyridoxine for PDE, biotin for biotinidase deficiency, BH4/sapropterin for responsive PKU and some neurotransmitter disorders, riboflavin for multiple acyl-CoA dehydrogenase deficiency. A leaky enzyme whose residual activity can be driven by saturating its cofactor is the reason a simple vitamin can be curative.

Replace the enzyme — and the blood–brain-barrier problem. Enzyme replacement therapy (ERT) delivers recombinant enzyme intravenously (agalsidase for Fabry, alglucosidase for Pompe, laronidase for MPS I, idursulfase for MPS II). It works well for systemic disease but is the clearest illustration of why neurometabolic disease is hard to treat: a ~60–80 kDa protein does not cross the blood–brain barrier, so IV ERT reaches liver, spleen, and marrow but largely spares the brain. This single pharmacokinetic fact is why intrathecal ERT, BBB-penetrant fusion constructs, and CNS-directed gene therapy are the active frontiers.

Reach the brain directly with gene therapy. Delivering a working gene to CNS tissue sidesteps both the barrier and the lifelong-adherence problem of diet. The paradigm shifts are already in clinic: onasemnogene abeparvovec (Zolgensma) for SMA (AAV9 crossing into the CNS), ex vivo lentiviral therapy for MLD (atidarsagene autotemcel — Libmeldy in the EU, Lenmeldy FDA-approved 2024), and AAV liver-directed trials in OTC deficiency and the organic acidemias. The organizing principle remains constant: identify whether you are fighting accumulation or deficiency, then choose the tool — and above all, intervene before the brain injury becomes structural, because in most IEM the neurological damage already laid down is not recoverable.